From 306f354c67397b3138300cde875c5cab45b857f7 Mon Sep 17 00:00:00 2001 From: Leon Romanovsky Date: Mon, 16 Mar 2020 09:31:03 +0200 Subject: net/mlx5_core: Set IB capability mask1 to fix ib_srpt connection failure The cap_mask1 isn't protected by field_select and not listed among RW fields, but it is required to be written to properly initialize ports in IB virtualization mode. Link: https://lore.kernel.org/linux-rdma/88bab94d2fd72f3145835b4518bc63dda587add6.camel@redhat.com Fixes: ab118da4c10a ("net/mlx5: Don't write read-only fields in MODIFY_HCA_VPORT_CONTEXT command") Signed-off-by: Leon Romanovsky Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/vport.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/vport.c b/drivers/net/ethernet/mellanox/mlx5/core/vport.c index 1faac31f74d0..23f879da9104 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/vport.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/vport.c @@ -1071,6 +1071,9 @@ int mlx5_core_modify_hca_vport_context(struct mlx5_core_dev *dev, MLX5_SET64(hca_vport_context, ctx, port_guid, req->port_guid); if (req->field_select & MLX5_HCA_VPORT_SEL_NODE_GUID) MLX5_SET64(hca_vport_context, ctx, node_guid, req->node_guid); + MLX5_SET(hca_vport_context, ctx, cap_mask1, req->cap_mask1); + MLX5_SET(hca_vport_context, ctx, cap_mask1_field_select, + req->cap_mask1_perm); err = mlx5_cmd_exec(dev, in, in_sz, out, sizeof(out)); ex: kfree(in); -- cgit v1.2.3 From 1de0306c3a05d305e45b1f1fabe2f4e94222eb6b Mon Sep 17 00:00:00 2001 From: Aya Levin Date: Mon, 9 Mar 2020 09:44:18 +0200 Subject: net/mlx5e: Enhance ICOSQ WQE info fields Add number of WQEBBs (WQE's Basic Block) to WQE info struct. Set the number of WQEBBs on WQE post, and increment the consumer counter (cc) on completion. In case of error completions, the cc was mistakenly not incremented, keeping a gap between cc and pc (producer counter). This failed the recovery flow on the ICOSQ from a CQE error which timed-out waiting for the cc and pc to meet. Fixes: be5323c8379f ("net/mlx5e: Report and recover from CQE error on ICOSQ") Signed-off-by: Aya Levin Reviewed-by: Tariq Toukan Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en.h | 1 + drivers/net/ethernet/mellanox/mlx5/core/en_rx.c | 11 +++++------ drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c | 1 + 3 files changed, 7 insertions(+), 6 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h index 220ef9f06f84..571ac5976549 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h @@ -371,6 +371,7 @@ enum { struct mlx5e_sq_wqe_info { u8 opcode; + u8 num_wqebbs; /* Auxiliary data for different opcodes. */ union { diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c index 1c3ab69cbd96..312d4692425b 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c @@ -477,6 +477,7 @@ static inline void mlx5e_fill_icosq_frag_edge(struct mlx5e_icosq *sq, /* fill sq frag edge with nops to avoid wqe wrapping two pages */ for (; wi < edge_wi; wi++) { wi->opcode = MLX5_OPCODE_NOP; + wi->num_wqebbs = 1; mlx5e_post_nop(wq, sq->sqn, &sq->pc); } } @@ -525,6 +526,7 @@ static int mlx5e_alloc_rx_mpwqe(struct mlx5e_rq *rq, u16 ix) umr_wqe->uctrl.xlt_offset = cpu_to_be16(xlt_offset); sq->db.ico_wqe[pi].opcode = MLX5_OPCODE_UMR; + sq->db.ico_wqe[pi].num_wqebbs = MLX5E_UMR_WQEBBS; sq->db.ico_wqe[pi].umr.rq = rq; sq->pc += MLX5E_UMR_WQEBBS; @@ -621,6 +623,7 @@ void mlx5e_poll_ico_cq(struct mlx5e_cq *cq) ci = mlx5_wq_cyc_ctr2ix(&sq->wq, sqcc); wi = &sq->db.ico_wqe[ci]; + sqcc += wi->num_wqebbs; if (last_wqe && unlikely(get_cqe_opcode(cqe) != MLX5_CQE_REQ)) { netdev_WARN_ONCE(cq->channel->netdev, @@ -631,16 +634,12 @@ void mlx5e_poll_ico_cq(struct mlx5e_cq *cq) break; } - if (likely(wi->opcode == MLX5_OPCODE_UMR)) { - sqcc += MLX5E_UMR_WQEBBS; + if (likely(wi->opcode == MLX5_OPCODE_UMR)) wi->umr.rq->mpwqe.umr_completed++; - } else if (likely(wi->opcode == MLX5_OPCODE_NOP)) { - sqcc++; - } else { + else if (unlikely(wi->opcode != MLX5_OPCODE_NOP)) netdev_WARN_ONCE(cq->channel->netdev, "Bad OPCODE in ICOSQ WQE info: 0x%x\n", wi->opcode); - } } while (!last_wqe); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c index 257a7c9f7a14..800d34ed8a96 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c @@ -78,6 +78,7 @@ void mlx5e_trigger_irq(struct mlx5e_icosq *sq) u16 pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc); sq->db.ico_wqe[pi].opcode = MLX5_OPCODE_NOP; + sq->db.ico_wqe[pi].num_wqebbs = 1; nopwqe = mlx5e_post_nop(wq, sq->sqn, &sq->pc); mlx5e_notify_hw(wq, sq->pc, sq->uar_map, &nopwqe->ctrl); } -- cgit v1.2.3 From 39369fd536d485a99a59d8e357c0d4d3ce19a3b8 Mon Sep 17 00:00:00 2001 From: Aya Levin Date: Thu, 12 Mar 2020 12:35:32 +0200 Subject: net/mlx5e: Fix missing reset of SW metadata in Striding RQ reset When resetting the RQ (moving RQ state from RST to RDY), the driver resets the WQ's SW metadata. In striding RQ mode, we maintain a field that reflects the actual expected WQ head (including in progress WQEs posted to the ICOSQ). It was mistakenly not reset together with the WQ. Fix this here. Fixes: 8276ea1353a4 ("net/mlx5e: Report and recover from CQE with error on RQ") Signed-off-by: Aya Levin Reviewed-by: Tariq Toukan Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h b/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h index a226277b0980..f07b1399744e 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h @@ -181,10 +181,12 @@ mlx5e_tx_dma_unmap(struct device *pdev, struct mlx5e_sq_dma *dma) static inline void mlx5e_rqwq_reset(struct mlx5e_rq *rq) { - if (rq->wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ) + if (rq->wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ) { mlx5_wq_ll_reset(&rq->mpwqe.wq); - else + rq->mpwqe.actual_wq_head = 0; + } else { mlx5_wq_cyc_reset(&rq->wqe.wq); + } } /* SW parser related functions */ -- cgit v1.2.3 From e239c6d686e1c37fb2ab143162dfb57471a8643f Mon Sep 17 00:00:00 2001 From: Aya Levin Date: Mon, 16 Mar 2020 16:53:10 +0200 Subject: net/mlx5e: Fix ICOSQ recovery flow with Striding RQ In striding RQ mode, the buffers of an RX WQE are first prepared and posted to the HW using a UMR WQEs via the ICOSQ. We maintain the state of these in-progress WQEs in the RQ SW struct. In the flow of ICOSQ recovery, the corresponding RQ is not in error state, hence: - The buffers of the in-progress WQEs must be released and the RQ metadata should reflect it. - Existing RX WQEs in the RQ should not be affected. For this, wrap the dealloc of the in-progress WQEs in a function, and use it in the ICOSQ recovery flow instead of mlx5e_free_rx_descs(). Fixes: be5323c8379f ("net/mlx5e: Report and recover from CQE error on ICOSQ") Signed-off-by: Aya Levin Reviewed-by: Tariq Toukan Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en.h | 1 + .../ethernet/mellanox/mlx5/core/en/reporter_rx.c | 2 +- drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 31 +++++++++++++++++----- 3 files changed, 26 insertions(+), 8 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h index 571ac5976549..c9606b8ab6ef 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h @@ -1060,6 +1060,7 @@ int mlx5e_modify_rq_state(struct mlx5e_rq *rq, int curr_state, int next_state); void mlx5e_activate_rq(struct mlx5e_rq *rq); void mlx5e_deactivate_rq(struct mlx5e_rq *rq); void mlx5e_free_rx_descs(struct mlx5e_rq *rq); +void mlx5e_free_rx_in_progress_descs(struct mlx5e_rq *rq); void mlx5e_activate_icosq(struct mlx5e_icosq *icosq); void mlx5e_deactivate_icosq(struct mlx5e_icosq *icosq); 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 6c72b592315b..a01e2de2488f 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c @@ -90,7 +90,7 @@ static int mlx5e_rx_reporter_err_icosq_cqe_recover(void *ctx) goto out; mlx5e_reset_icosq_cc_pc(icosq); - mlx5e_free_rx_descs(rq); + mlx5e_free_rx_in_progress_descs(rq); clear_bit(MLX5E_SQ_STATE_RECOVERING, &icosq->state); mlx5e_activate_icosq(icosq); mlx5e_activate_rq(rq); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c index 21de4764d4c0..4ef3dc79f73c 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c @@ -813,6 +813,29 @@ int mlx5e_wait_for_min_rx_wqes(struct mlx5e_rq *rq, int wait_time) return -ETIMEDOUT; } +void mlx5e_free_rx_in_progress_descs(struct mlx5e_rq *rq) +{ + struct mlx5_wq_ll *wq; + u16 head; + int i; + + if (rq->wq_type != MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ) + return; + + wq = &rq->mpwqe.wq; + head = wq->head; + + /* Outstanding UMR WQEs (in progress) start at wq->head */ + for (i = 0; i < rq->mpwqe.umr_in_progress; i++) { + rq->dealloc_wqe(rq, head); + head = mlx5_wq_ll_get_wqe_next_ix(wq, head); + } + + rq->mpwqe.actual_wq_head = wq->head; + rq->mpwqe.umr_in_progress = 0; + rq->mpwqe.umr_completed = 0; +} + void mlx5e_free_rx_descs(struct mlx5e_rq *rq) { __be16 wqe_ix_be; @@ -820,14 +843,8 @@ void mlx5e_free_rx_descs(struct mlx5e_rq *rq) if (rq->wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ) { struct mlx5_wq_ll *wq = &rq->mpwqe.wq; - u16 head = wq->head; - int i; - /* Outstanding UMR WQEs (in progress) start at wq->head */ - for (i = 0; i < rq->mpwqe.umr_in_progress; i++) { - rq->dealloc_wqe(rq, head); - head = mlx5_wq_ll_get_wqe_next_ix(wq, head); - } + mlx5e_free_rx_in_progress_descs(rq); while (!mlx5_wq_ll_is_empty(wq)) { struct mlx5e_rx_wqe_ll *wqe; -- cgit v1.2.3 From 187a9830c921d92c4a9a8e2921ecc4b35a97532c Mon Sep 17 00:00:00 2001 From: Aya Levin Date: Thu, 19 Mar 2020 13:25:17 +0200 Subject: net/mlx5e: Do not recover from a non-fatal syndrome For non-fatal syndromes like LOCAL_LENGTH_ERR, recovery shouldn't be triggered. In these scenarios, the RQ is not actually in ERR state. This misleads the recovery flow which assumes that the RQ is really in error state and no more completions arrive, causing crashes on bad page state. Fixes: 8276ea1353a4 ("net/mlx5e: Report and recover from CQE with error on RQ") Signed-off-by: Aya Levin Reviewed-by: Tariq Toukan Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en/health.h | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/health.h b/drivers/net/ethernet/mellanox/mlx5/core/en/health.h index d3693fa547ac..e54f70d9af22 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/health.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/health.h @@ -10,8 +10,7 @@ static inline bool cqe_syndrome_needs_recover(u8 syndrome) { - return syndrome == MLX5_CQE_SYNDROME_LOCAL_LENGTH_ERR || - syndrome == MLX5_CQE_SYNDROME_LOCAL_QP_OP_ERR || + return syndrome == MLX5_CQE_SYNDROME_LOCAL_QP_OP_ERR || syndrome == MLX5_CQE_SYNDROME_LOCAL_PROT_ERR || syndrome == MLX5_CQE_SYNDROME_WR_FLUSH_ERR; } -- cgit v1.2.3