From ac77998b7ac3044f0509b097da9637184598980d Mon Sep 17 00:00:00 2001 From: Mohammad Kabat Date: Thu, 25 Mar 2021 14:38:55 +0200 Subject: [PATCH 1/5] net/mlx5: Fix size field in bufferx_reg struct According to HW spec the field "size" should be 16 bits in bufferx register. Fixes: e281682bf294 ("net/mlx5_core: HW data structs/types definitions cleanup") Signed-off-by: Mohammad Kabat Reviewed-by: Moshe Shemesh Signed-off-by: Saeed Mahameed --- include/linux/mlx5/mlx5_ifc.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h index 598ac3bcc901..5743f5b3414b 100644 --- a/include/linux/mlx5/mlx5_ifc.h +++ b/include/linux/mlx5/mlx5_ifc.h @@ -9900,8 +9900,8 @@ struct mlx5_ifc_bufferx_reg_bits { u8 reserved_at_0[0x6]; u8 lossy[0x1]; u8 epsb[0x1]; - u8 reserved_at_8[0xc]; - u8 size[0xc]; + u8 reserved_at_8[0x8]; + u8 size[0x10]; u8 xoff_threshold[0x10]; u8 xon_threshold[0x10]; From 063bd355595428750803d8736a9bb7c8db67d42d Mon Sep 17 00:00:00 2001 From: Moshe Shemesh Date: Fri, 4 Feb 2022 11:47:44 +0200 Subject: [PATCH 2/5] net/mlx5: Fix a race on command flush flow Fix a refcount use after free warning due to a race on command entry. Such race occurs when one of the commands releases its last refcount and frees its index and entry while another process running command flush flow takes refcount to this command entry. The process which handles commands flush may see this command as needed to be flushed if the other process released its refcount but didn't release the index yet. Fix it by adding the needed spin lock. It fixes the following warning trace: refcount_t: addition on 0; use-after-free. WARNING: CPU: 11 PID: 540311 at lib/refcount.c:25 refcount_warn_saturate+0x80/0xe0 ... RIP: 0010:refcount_warn_saturate+0x80/0xe0 ... Call Trace: mlx5_cmd_trigger_completions+0x293/0x340 [mlx5_core] mlx5_cmd_flush+0x3a/0xf0 [mlx5_core] enter_error_state+0x44/0x80 [mlx5_core] mlx5_fw_fatal_reporter_err_work+0x37/0xe0 [mlx5_core] process_one_work+0x1be/0x390 worker_thread+0x4d/0x3d0 ? rescuer_thread+0x350/0x350 kthread+0x141/0x160 ? set_kthread_struct+0x40/0x40 ret_from_fork+0x1f/0x30 Fixes: 50b2412b7e78 ("net/mlx5: Avoid possible free of command entry while timeout comp handler") Signed-off-by: Moshe Shemesh Reviewed-by: Eran Ben Elisha Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/cmd.c | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c index 17fe05809653..3eacd8739929 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c @@ -131,11 +131,8 @@ static int cmd_alloc_index(struct mlx5_cmd *cmd) static void cmd_free_index(struct mlx5_cmd *cmd, int idx) { - unsigned long flags; - - spin_lock_irqsave(&cmd->alloc_lock, flags); + lockdep_assert_held(&cmd->alloc_lock); set_bit(idx, &cmd->bitmask); - spin_unlock_irqrestore(&cmd->alloc_lock, flags); } static void cmd_ent_get(struct mlx5_cmd_work_ent *ent) @@ -145,17 +142,21 @@ static void cmd_ent_get(struct mlx5_cmd_work_ent *ent) static void cmd_ent_put(struct mlx5_cmd_work_ent *ent) { + struct mlx5_cmd *cmd = ent->cmd; + unsigned long flags; + + spin_lock_irqsave(&cmd->alloc_lock, flags); if (!refcount_dec_and_test(&ent->refcnt)) - return; + goto out; if (ent->idx >= 0) { - struct mlx5_cmd *cmd = ent->cmd; - cmd_free_index(cmd, ent->idx); up(ent->page_queue ? &cmd->pages_sem : &cmd->sem); } cmd_free_ent(ent); +out: + spin_unlock_irqrestore(&cmd->alloc_lock, flags); } static struct mlx5_cmd_layout *get_inst(struct mlx5_cmd *cmd, int idx) From 39bab83b119faac4bf7f07173a42ed35be95147e Mon Sep 17 00:00:00 2001 From: Dima Chumak Date: Mon, 17 Jan 2022 15:32:16 +0200 Subject: [PATCH 3/5] net/mlx5: Fix offloading with ESWITCH_IPV4_TTL_MODIFY_ENABLE Only prio 1 is supported for nic mode when there is no ignore flow level support in firmware. But for switchdev mode, which supports fixed number of statically pre-allocated prios, this restriction is not relevant so it can be relaxed. Fixes: d671e109bd85 ("net/mlx5: Fix tc max supported prio for nic mode") Signed-off-by: Dima Chumak Reviewed-by: Roi Dayan Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/lib/fs_chains.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_chains.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_chains.c index 1e8ec4f236b2..df58cba37930 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_chains.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_chains.c @@ -121,9 +121,6 @@ u32 mlx5_chains_get_nf_ft_chain(struct mlx5_fs_chains *chains) u32 mlx5_chains_get_prio_range(struct mlx5_fs_chains *chains) { - if (!mlx5_chains_prios_supported(chains)) - return 1; - if (mlx5_chains_ignore_flow_level_supported(chains)) return UINT_MAX; From ad11c4f1d8fd1f03639460e425a36f7fd0ea83f5 Mon Sep 17 00:00:00 2001 From: Roi Dayan Date: Wed, 16 Feb 2022 13:56:57 +0200 Subject: [PATCH 4/5] net/mlx5e: Lag, Only handle events from highest priority multipath entry There could be multiple multipath entries but changing the port affinity for each one doesn't make much sense and there should be a default one. So only track the entry with lowest priority value. The commit doesn't affect existing users with a single entry. Fixes: 544fe7c2e654 ("net/mlx5e: Activate HW multipath and handle port affinity based on FIB events") Signed-off-by: Roi Dayan Reviewed-by: Maor Dickman Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c index 1ca01a5b6cdd..626aa60b6099 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c @@ -126,6 +126,10 @@ static void mlx5_lag_fib_route_event(struct mlx5_lag *ldev, return; } + /* Handle multipath entry with lower priority value */ + if (mp->mfi && mp->mfi != fi && fi->fib_priority >= mp->mfi->fib_priority) + return; + /* Handle add/replace event */ nhs = fib_info_num_path(fi); if (nhs == 1) { @@ -135,12 +139,13 @@ static void mlx5_lag_fib_route_event(struct mlx5_lag *ldev, int i = mlx5_lag_dev_get_netdev_idx(ldev, nh_dev); if (i < 0) - i = MLX5_LAG_NORMAL_AFFINITY; - else - ++i; + return; + i++; mlx5_lag_set_port_affinity(ldev, i); } + + mp->mfi = fi; return; } From 99a2b9be077ae3a5d97fbf5f7782e0f2e9812978 Mon Sep 17 00:00:00 2001 From: Ben Ben-Ishay Date: Wed, 2 Mar 2022 17:07:08 +0200 Subject: [PATCH 5/5] net/mlx5e: SHAMPO, reduce TIR indication SHAMPO is an RQ / WQ feature, an indication was added to the TIR in the first place to enforce suitability between connected TIR and RQ, this enforcement does not exist in current the Firmware implementation and was redundant in the first place. Fixes: 83439f3c37aa ("net/mlx5e: Add HW-GRO offload") Signed-off-by: Ben Ben-Ishay Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en/tir.c | 3 --- drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 3 +-- include/linux/mlx5/mlx5_ifc.h | 1 - 3 files changed, 1 insertion(+), 6 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tir.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tir.c index da169b816665..d4239e3b3c88 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/tir.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tir.c @@ -88,9 +88,6 @@ void mlx5e_tir_builder_build_packet_merge(struct mlx5e_tir_builder *builder, (MLX5E_PARAMS_DEFAULT_LRO_WQE_SZ - rough_max_l2_l3_hdr_sz) >> 8); MLX5_SET(tirc, tirc, lro_timeout_period_usecs, pkt_merge_param->timeout); break; - case MLX5E_PACKET_MERGE_SHAMPO: - MLX5_SET(tirc, tirc, packet_merge_mask, MLX5_TIRC_PACKET_MERGE_MASK_SHAMPO); - break; default: break; } diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c index bf80fb612449..3667f5ef5990 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c @@ -3616,8 +3616,7 @@ static int set_feature_hw_gro(struct net_device *netdev, bool enable) goto out; } - err = mlx5e_safe_switch_params(priv, &new_params, - mlx5e_modify_tirs_packet_merge_ctx, NULL, reset); + err = mlx5e_safe_switch_params(priv, &new_params, NULL, NULL, reset); out: mutex_unlock(&priv->state_lock); return err; diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h index 5743f5b3414b..49a48d7709ac 100644 --- a/include/linux/mlx5/mlx5_ifc.h +++ b/include/linux/mlx5/mlx5_ifc.h @@ -3434,7 +3434,6 @@ enum { enum { MLX5_TIRC_PACKET_MERGE_MASK_IPV4_LRO = BIT(0), MLX5_TIRC_PACKET_MERGE_MASK_IPV6_LRO = BIT(1), - MLX5_TIRC_PACKET_MERGE_MASK_SHAMPO = BIT(2), }; enum {