diff options
30 files changed, 832 insertions, 156 deletions
diff --git a/drivers/infiniband/core/iwcm.c b/drivers/infiniband/core/iwcm.c index 0301fcad4b48..1a6339f3a63f 100644 --- a/drivers/infiniband/core/iwcm.c +++ b/drivers/infiniband/core/iwcm.c @@ -143,8 +143,8 @@ static struct iwcm_work *get_work(struct iwcm_id_private *cm_id_priv) if (list_empty(&cm_id_priv->work_free_list)) return NULL; - work = list_entry(cm_id_priv->work_free_list.next, struct iwcm_work, - free_list); + work = list_first_entry(&cm_id_priv->work_free_list, struct iwcm_work, + free_list); list_del_init(&work->free_list); return work; } @@ -206,17 +206,17 @@ static void free_cm_id(struct iwcm_id_private *cm_id_priv) /* * Release a reference on cm_id. If the last reference is being - * released, free the cm_id and return 1. + * released, free the cm_id and return 'true'. */ -static int iwcm_deref_id(struct iwcm_id_private *cm_id_priv) +static bool iwcm_deref_id(struct iwcm_id_private *cm_id_priv) { if (refcount_dec_and_test(&cm_id_priv->refcount)) { BUG_ON(!list_empty(&cm_id_priv->work_list)); free_cm_id(cm_id_priv); - return 1; + return true; } - return 0; + return false; } static void add_ref(struct iw_cm_id *cm_id) @@ -368,8 +368,10 @@ EXPORT_SYMBOL(iw_cm_disconnect); * * Clean up all resources associated with the connection and release * the initial reference taken by iw_create_cm_id. + * + * Returns true if and only if the last cm_id_priv reference has been dropped. */ -static void destroy_cm_id(struct iw_cm_id *cm_id) +static bool destroy_cm_id(struct iw_cm_id *cm_id) { struct iwcm_id_private *cm_id_priv; struct ib_qp *qp; @@ -439,7 +441,7 @@ static void destroy_cm_id(struct iw_cm_id *cm_id) iwpm_remove_mapping(&cm_id->local_addr, RDMA_NL_IWCM); } - (void)iwcm_deref_id(cm_id_priv); + return iwcm_deref_id(cm_id_priv); } /* @@ -450,7 +452,8 @@ static void destroy_cm_id(struct iw_cm_id *cm_id) */ void iw_destroy_cm_id(struct iw_cm_id *cm_id) { - destroy_cm_id(cm_id); + if (!destroy_cm_id(cm_id)) + flush_workqueue(iwcm_wq); } EXPORT_SYMBOL(iw_destroy_cm_id); @@ -1017,16 +1020,13 @@ static void cm_work_handler(struct work_struct *_work) struct iw_cm_event levent; struct iwcm_id_private *cm_id_priv = work->cm_id; unsigned long flags; - int empty; int ret = 0; spin_lock_irqsave(&cm_id_priv->lock, flags); - empty = list_empty(&cm_id_priv->work_list); - while (!empty) { - work = list_entry(cm_id_priv->work_list.next, - struct iwcm_work, list); + while (!list_empty(&cm_id_priv->work_list)) { + work = list_first_entry(&cm_id_priv->work_list, + struct iwcm_work, list); list_del_init(&work->list); - empty = list_empty(&cm_id_priv->work_list); levent = work->event; put_work(work); spin_unlock_irqrestore(&cm_id_priv->lock, flags); @@ -1034,13 +1034,11 @@ static void cm_work_handler(struct work_struct *_work) if (!test_bit(IWCM_F_DROP_EVENTS, &cm_id_priv->flags)) { ret = process_event(cm_id_priv, &levent); if (ret) - destroy_cm_id(&cm_id_priv->id); + WARN_ON_ONCE(destroy_cm_id(&cm_id_priv->id)); } else pr_debug("dropping event %d\n", levent.event); if (iwcm_deref_id(cm_id_priv)) return; - if (empty) - return; spin_lock_irqsave(&cm_id_priv->lock, flags); } spin_unlock_irqrestore(&cm_id_priv->lock, flags); @@ -1093,11 +1091,8 @@ static int cm_event_handler(struct iw_cm_id *cm_id, } refcount_inc(&cm_id_priv->refcount); - if (list_empty(&cm_id_priv->work_list)) { - list_add_tail(&work->list, &cm_id_priv->work_list); - queue_work(iwcm_wq, &work->work); - } else - list_add_tail(&work->list, &cm_id_priv->work_list); + list_add_tail(&work->list, &cm_id_priv->work_list); + queue_work(iwcm_wq, &work->work); out: spin_unlock_irqrestore(&cm_id_priv->lock, flags); return ret; diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c index ce9c5bae83bf..d261b09025ca 100644 --- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c +++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c @@ -4201,6 +4201,9 @@ int bnxt_re_alloc_ucontext(struct ib_ucontext *ctx, struct ib_udata *udata) if (rdev->pacing.dbr_pacing) resp.comp_mask |= BNXT_RE_UCNTX_CMASK_DBR_PACING_ENABLED; + if (_is_host_msn_table(rdev->qplib_res.dattr->dev_cap_flags2)) + resp.comp_mask |= BNXT_RE_UCNTX_CMASK_MSN_TABLE_ENABLED; + if (udata->inlen >= sizeof(ureq)) { rc = ib_copy_from_udata(&ureq, udata, min(udata->inlen, sizeof(ureq))); if (rc) diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.c b/drivers/infiniband/hw/bnxt_re/qplib_fp.c index 04258676d072..49e4a4a50bfa 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_fp.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.c @@ -984,7 +984,7 @@ int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) u16 nsge; if (res->dattr) - qp->dev_cap_flags = res->dattr->dev_cap_flags; + qp->is_host_msn_tbl = _is_host_msn_table(res->dattr->dev_cap_flags2); sq->dbinfo.flags = 0; bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req, @@ -1002,7 +1002,7 @@ int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) sizeof(struct sq_psn_search_ext) : sizeof(struct sq_psn_search); - if (BNXT_RE_HW_RETX(qp->dev_cap_flags)) { + if (qp->is_host_msn_tbl) { psn_sz = sizeof(struct sq_msn_search); qp->msn = 0; } @@ -1016,7 +1016,7 @@ int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) hwq_attr.aux_depth = psn_sz ? bnxt_qplib_set_sq_size(sq, qp->wqe_mode) : 0; /* Update msn tbl size */ - if (BNXT_RE_HW_RETX(qp->dev_cap_flags) && psn_sz) { + if (qp->is_host_msn_tbl && psn_sz) { hwq_attr.aux_depth = roundup_pow_of_two(bnxt_qplib_set_sq_size(sq, qp->wqe_mode)); qp->msn_tbl_sz = hwq_attr.aux_depth; qp->msn = 0; @@ -1637,7 +1637,7 @@ static void bnxt_qplib_fill_psn_search(struct bnxt_qplib_qp *qp, if (!swq->psn_search) return; /* Handle MSN differently on cap flags */ - if (BNXT_RE_HW_RETX(qp->dev_cap_flags)) { + if (qp->is_host_msn_tbl) { bnxt_qplib_fill_msn_search(qp, wqe, swq); return; } @@ -1819,7 +1819,7 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp, } swq = bnxt_qplib_get_swqe(sq, &wqe_idx); - bnxt_qplib_pull_psn_buff(qp, sq, swq, BNXT_RE_HW_RETX(qp->dev_cap_flags)); + bnxt_qplib_pull_psn_buff(qp, sq, swq, qp->is_host_msn_tbl); idx = 0; swq->slot_idx = hwq->prod; @@ -2009,7 +2009,7 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp, rc = -EINVAL; goto done; } - if (!BNXT_RE_HW_RETX(qp->dev_cap_flags) || msn_update) { + if (!qp->is_host_msn_tbl || msn_update) { swq->next_psn = sq->psn & BTH_PSN_MASK; bnxt_qplib_fill_psn_search(qp, wqe, swq); } diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.h b/drivers/infiniband/hw/bnxt_re/qplib_fp.h index 7fd4506b3584..4aaac84c1b1b 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_fp.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.h @@ -340,7 +340,7 @@ struct bnxt_qplib_qp { struct list_head rq_flush; u32 msn; u32 msn_tbl_sz; - u16 dev_cap_flags; + bool is_host_msn_tbl; }; #define BNXT_QPLIB_MAX_CQE_ENTRY_SIZE sizeof(struct cq_base) diff --git a/drivers/infiniband/hw/bnxt_re/qplib_res.h b/drivers/infiniband/hw/bnxt_re/qplib_res.h index 61628f7f1253..a0f78cde314f 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_res.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_res.h @@ -554,6 +554,12 @@ static inline bool _is_hw_retx_supported(u16 dev_cap_flags) #define BNXT_RE_HW_RETX(a) _is_hw_retx_supported((a)) +static inline bool _is_host_msn_table(u16 dev_cap_ext_flags2) +{ + return (dev_cap_ext_flags2 & CREQ_QUERY_FUNC_RESP_SB_REQ_RETRANSMISSION_SUPPORT_MASK) == + CREQ_QUERY_FUNC_RESP_SB_REQ_RETRANSMISSION_SUPPORT_HOST_MSN_TABLE; +} + static inline u8 bnxt_qplib_dbr_pacing_en(struct bnxt_qplib_chip_ctx *cctx) { return cctx->modes.dbr_pacing; diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.c b/drivers/infiniband/hw/bnxt_re/qplib_sp.c index 8beeedd15061..9328db92fa6d 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_sp.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.c @@ -156,6 +156,7 @@ int bnxt_qplib_get_dev_attr(struct bnxt_qplib_rcfw *rcfw, (0x01 << RCFW_DBR_BASE_PAGE_SHIFT); attr->max_sgid = BNXT_QPLIB_NUM_GIDS_SUPPORTED; attr->dev_cap_flags = le16_to_cpu(sb->dev_cap_flags); + attr->dev_cap_flags2 = le16_to_cpu(sb->dev_cap_ext_flags_2); bnxt_qplib_query_version(rcfw, attr->fw_ver); diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.h b/drivers/infiniband/hw/bnxt_re/qplib_sp.h index d33c78b96217..16a67d70a6fc 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_sp.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.h @@ -72,6 +72,7 @@ struct bnxt_qplib_dev_attr { u8 tqm_alloc_reqs[MAX_TQM_ALLOC_REQ]; bool is_atomic; u16 dev_cap_flags; + u16 dev_cap_flags2; u32 max_dpi; }; diff --git a/drivers/infiniband/hw/bnxt_re/roce_hsi.h b/drivers/infiniband/hw/bnxt_re/roce_hsi.h index 605c9463c408..042530969505 100644 --- a/drivers/infiniband/hw/bnxt_re/roce_hsi.h +++ b/drivers/infiniband/hw/bnxt_re/roce_hsi.h @@ -2157,8 +2157,36 @@ struct creq_query_func_resp_sb { __le32 tqm_alloc_reqs[12]; __le32 max_dpi; u8 max_sge_var_wqe; - u8 reserved_8; + u8 dev_cap_ext_flags; + #define CREQ_QUERY_FUNC_RESP_SB_ATOMIC_OPS_NOT_SUPPORTED 0x1UL + #define CREQ_QUERY_FUNC_RESP_SB_DRV_VERSION_RGTR_SUPPORTED 0x2UL + #define CREQ_QUERY_FUNC_RESP_SB_CREATE_QP_BATCH_SUPPORTED 0x4UL + #define CREQ_QUERY_FUNC_RESP_SB_DESTROY_QP_BATCH_SUPPORTED 0x8UL + #define CREQ_QUERY_FUNC_RESP_SB_ROCE_STATS_EXT_CTX_SUPPORTED 0x10UL + #define CREQ_QUERY_FUNC_RESP_SB_CREATE_SRQ_SGE_SUPPORTED 0x20UL + #define CREQ_QUERY_FUNC_RESP_SB_FIXED_SIZE_WQE_DISABLED 0x40UL + #define CREQ_QUERY_FUNC_RESP_SB_DCN_SUPPORTED 0x80UL __le16 max_inline_data_var_wqe; + __le32 start_qid; + u8 max_msn_table_size; + u8 reserved8_1; + __le16 dev_cap_ext_flags_2; + #define CREQ_QUERY_FUNC_RESP_SB_OPTIMIZE_MODIFY_QP_SUPPORTED 0x1UL + #define CREQ_QUERY_FUNC_RESP_SB_CHANGE_UDP_SRC_PORT_WQE_SUPPORTED 0x2UL + #define CREQ_QUERY_FUNC_RESP_SB_CQ_COALESCING_SUPPORTED 0x4UL + #define CREQ_QUERY_FUNC_RESP_SB_MEMORY_REGION_RO_SUPPORTED 0x8UL + #define CREQ_QUERY_FUNC_RESP_SB_REQ_RETRANSMISSION_SUPPORT_MASK 0x30UL + #define CREQ_QUERY_FUNC_RESP_SB_REQ_RETRANSMISSION_SUPPORT_SFT 4 + #define CREQ_QUERY_FUNC_RESP_SB_REQ_RETRANSMISSION_SUPPORT_HOST_PSN_TABLE (0x0UL << 4) + #define CREQ_QUERY_FUNC_RESP_SB_REQ_RETRANSMISSION_SUPPORT_HOST_MSN_TABLE (0x1UL << 4) + #define CREQ_QUERY_FUNC_RESP_SB_REQ_RETRANSMISSION_SUPPORT_IQM_MSN_TABLE (0x2UL << 4) + #define CREQ_QUERY_FUNC_RESP_SB_REQ_RETRANSMISSION_SUPPORT_LAST \ + CREQ_QUERY_FUNC_RESP_SB_REQ_RETRANSMISSION_SUPPORT_IQM_MSN_TABLE + __le16 max_xp_qp_size; + __le16 create_qp_batch_size; + __le16 destroy_qp_batch_size; + __le16 reserved16; + __le64 reserved64; }; /* cmdq_set_func_resources (size:448b/56B) */ diff --git a/drivers/infiniband/hw/efa/efa_com.c b/drivers/infiniband/hw/efa/efa_com.c index 16a24a05fc2a..bafd210dd43e 100644 --- a/drivers/infiniband/hw/efa/efa_com.c +++ b/drivers/infiniband/hw/efa/efa_com.c @@ -1,6 +1,6 @@ // SPDX-License-Identifier: GPL-2.0 OR BSD-2-Clause /* - * Copyright 2018-2021 Amazon.com, Inc. or its affiliates. All rights reserved. + * Copyright 2018-2024 Amazon.com, Inc. or its affiliates. All rights reserved. */ #include "efa_com.h" @@ -406,8 +406,8 @@ static struct efa_comp_ctx *efa_com_submit_admin_cmd(struct efa_com_admin_queue return comp_ctx; } -static void efa_com_handle_single_admin_completion(struct efa_com_admin_queue *aq, - struct efa_admin_acq_entry *cqe) +static int efa_com_handle_single_admin_completion(struct efa_com_admin_queue *aq, + struct efa_admin_acq_entry *cqe) { struct efa_comp_ctx *comp_ctx; u16 cmd_id; @@ -416,11 +416,11 @@ static void efa_com_handle_single_admin_completion(struct efa_com_admin_queue *a EFA_ADMIN_ACQ_COMMON_DESC_COMMAND_ID); comp_ctx = efa_com_get_comp_ctx(aq, cmd_id, false); - if (!comp_ctx) { + if (comp_ctx->status != EFA_CMD_SUBMITTED) { ibdev_err(aq->efa_dev, - "comp_ctx is NULL. Changing the admin queue running state\n"); - clear_bit(EFA_AQ_STATE_RUNNING_BIT, &aq->state); - return; + "Received completion with unexpected command id[%d], sq producer: %d, sq consumer: %d, cq consumer: %d\n", + cmd_id, aq->sq.pc, aq->sq.cc, aq->cq.cc); + return -EINVAL; } comp_ctx->status = EFA_CMD_COMPLETED; @@ -428,14 +428,17 @@ static void efa_com_handle_single_admin_completion(struct efa_com_admin_queue *a if (!test_bit(EFA_AQ_STATE_POLLING_BIT, &aq->state)) complete(&comp_ctx->wait_event); + + return 0; } static void efa_com_handle_admin_completion(struct efa_com_admin_queue *aq) { struct efa_admin_acq_entry *cqe; u16 queue_size_mask; - u16 comp_num = 0; + u16 comp_cmds = 0; u8 phase; + int err; u16 ci; queue_size_mask = aq->depth - 1; @@ -453,10 +456,12 @@ static void efa_com_handle_admin_completion(struct efa_com_admin_queue *aq) * phase bit was validated */ dma_rmb(); - efa_com_handle_single_admin_completion(aq, cqe); + err = efa_com_handle_single_admin_completion(aq, cqe); + if (!err) + comp_cmds++; + aq->cq.cc++; ci++; - comp_num++; if (ci == aq->depth) { ci = 0; phase = !phase; @@ -465,10 +470,9 @@ static void efa_com_handle_admin_completion(struct efa_com_admin_queue *aq) cqe = &aq->cq.entries[ci]; } - aq->cq.cc += comp_num; aq->cq.phase = phase; - aq->sq.cc += comp_num; - atomic64_add(comp_num, &aq->stats.completed_cmd); + aq->sq.cc += comp_cmds; + atomic64_add(comp_cmds, &aq->stats.completed_cmd); } static int efa_com_comp_status_to_errno(u8 comp_status) diff --git a/drivers/infiniband/hw/efa/efa_main.c b/drivers/infiniband/hw/efa/efa_main.c index d1a48f988f6c..99f9ac23c721 100644 --- a/drivers/infiniband/hw/efa/efa_main.c +++ b/drivers/infiniband/hw/efa/efa_main.c @@ -190,15 +190,23 @@ static int efa_request_doorbell_bar(struct efa_dev *dev) { u8 db_bar_idx = dev->dev_attr.db_bar; struct pci_dev *pdev = dev->pdev; - int bars; + int pci_mem_bars; + int db_bar; int err; - if (!(BIT(db_bar_idx) & EFA_BASE_BAR_MASK)) { - bars = pci_select_bars(pdev, IORESOURCE_MEM) & BIT(db_bar_idx); + db_bar = BIT(db_bar_idx); + if (!(db_bar & EFA_BASE_BAR_MASK)) { + pci_mem_bars = pci_select_bars(pdev, IORESOURCE_MEM); + if (db_bar & ~pci_mem_bars) { + dev_err(&pdev->dev, + "Doorbells BAR unavailable. Requested %#x, available %#x\n", + db_bar, pci_mem_bars); + return -ENODEV; + } - err = pci_request_selected_regions(pdev, bars, DRV_MODULE_NAME); + err = pci_request_selected_regions(pdev, db_bar, DRV_MODULE_NAME); if (err) { - dev_err(&dev->pdev->dev, + dev_err(&pdev->dev, "pci_request_selected_regions for bar %d failed %d\n", db_bar_idx, err); return err; @@ -531,7 +539,7 @@ static struct efa_dev *efa_probe_device(struct pci_dev *pdev) { struct efa_com_dev *edev; struct efa_dev *dev; - int bars; + int pci_mem_bars; int err; err = pci_enable_device_mem(pdev); @@ -556,8 +564,14 @@ static struct efa_dev *efa_probe_device(struct pci_dev *pdev) dev->pdev = pdev; xa_init(&dev->cqs_xa); - bars = pci_select_bars(pdev, IORESOURCE_MEM) & EFA_BASE_BAR_MASK; - err = pci_request_selected_regions(pdev, bars, DRV_MODULE_NAME); + pci_mem_bars = pci_select_bars(pdev, IORESOURCE_MEM); + if (EFA_BASE_BAR_MASK & ~pci_mem_bars) { + dev_err(&pdev->dev, "BARs unavailable. Requested %#x, available %#x\n", + (int)EFA_BASE_BAR_MASK, pci_mem_bars); + err = -ENODEV; + goto err_ibdev_destroy; + } + err = pci_request_selected_regions(pdev, EFA_BASE_BAR_MASK, DRV_MODULE_NAME); if (err) { dev_err(&pdev->dev, "pci_request_selected_regions failed %d\n", err); diff --git a/drivers/infiniband/hw/irdma/main.h b/drivers/infiniband/hw/irdma/main.h index b65bc2ea542f..9f0ed6e84471 100644 --- a/drivers/infiniband/hw/irdma/main.h +++ b/drivers/infiniband/hw/irdma/main.h @@ -239,7 +239,7 @@ struct irdma_qv_info { struct irdma_qvlist_info { u32 num_vectors; - struct irdma_qv_info qv_info[]; + struct irdma_qv_info qv_info[] __counted_by(num_vectors); }; struct irdma_gen_ops { diff --git a/drivers/infiniband/hw/mana/device.c b/drivers/infiniband/hw/mana/device.c index 7e09ceb3da53..b07a8e2e838f 100644 --- a/drivers/infiniband/hw/mana/device.c +++ b/drivers/infiniband/hw/mana/device.c @@ -5,6 +5,7 @@ #include "mana_ib.h" #include <net/mana/mana_auxiliary.h> +#include <net/addrconf.h> MODULE_DESCRIPTION("Microsoft Azure Network Adapter IB driver"); MODULE_LICENSE("GPL"); @@ -92,6 +93,7 @@ static int mana_ib_probe(struct auxiliary_device *adev, goto free_ib_device; } ether_addr_copy(mac_addr, upper_ndev->dev_addr); + addrconf_addr_eui48((u8 *)&dev->ib_dev.node_guid, upper_ndev->dev_addr); ret = ib_device_set_netdev(&dev->ib_dev, upper_ndev, 1); rcu_read_unlock(); if (ret) { @@ -124,6 +126,7 @@ static int mana_ib_probe(struct auxiliary_device *adev, if (ret) goto destroy_eqs; + xa_init_flags(&dev->qp_table_wq, XA_FLAGS_LOCK_IRQ); ret = mana_ib_gd_config_mac(dev, ADDR_OP_ADD, mac_addr); if (ret) { ibdev_err(&dev->ib_dev, "Failed to add Mac address, ret %d", @@ -141,6 +144,7 @@ static int mana_ib_probe(struct auxiliary_device *adev, return 0; destroy_rnic: + xa_destroy(&dev->qp_table_wq); mana_ib_gd_destroy_rnic_adapter(dev); destroy_eqs: mana_ib_destroy_eqs(dev); @@ -156,6 +160,7 @@ static void mana_ib_remove(struct auxiliary_device *adev) struct mana_ib_dev *dev = dev_get_drvdata(&adev->dev); ib_unregister_device(&dev->ib_dev); + xa_destroy(&dev->qp_table_wq); mana_ib_gd_destroy_rnic_adapter(dev); mana_ib_destroy_eqs(dev); mana_gd_deregister_device(dev->gdma_dev); diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c index 2a411357640e..d13abc954d2a 100644 --- a/drivers/infiniband/hw/mana/main.c +++ b/drivers/infiniband/hw/mana/main.c @@ -547,14 +547,27 @@ int mana_ib_query_device(struct ib_device *ibdev, struct ib_device_attr *props, struct mana_ib_dev *dev = container_of(ibdev, struct mana_ib_dev, ib_dev); + memset(props, 0, sizeof(*props)); + props->max_mr_size = MANA_IB_MAX_MR_SIZE; + props->page_size_cap = PAGE_SZ_BM; props->max_qp = dev->adapter_caps.max_qp_count; props->max_qp_wr = dev->adapter_caps.max_qp_wr; + props->device_cap_flags = IB_DEVICE_RC_RNR_NAK_GEN; + props->max_send_sge = dev->adapter_caps.max_send_sge_count; + props->max_recv_sge = dev->adapter_caps.max_recv_sge_count; + props->max_sge_rd = dev->adapter_caps.max_recv_sge_count; props->max_cq = dev->adapter_caps.max_cq_count; props->max_cqe = dev->adapter_caps.max_qp_wr; props->max_mr = dev->adapter_caps.max_mr_count; - props->max_mr_size = MANA_IB_MAX_MR_SIZE; - props->max_send_sge = dev->adapter_caps.max_send_sge_count; - props->max_recv_sge = dev->adapter_caps.max_recv_sge_count; + props->max_pd = dev->adapter_caps.max_pd_count; + props->max_qp_rd_atom = dev->adapter_caps.max_inbound_read_limit; + props->max_res_rd_atom = props->max_qp_rd_atom * props->max_qp; + props->max_qp_init_rd_atom = dev->adapter_caps.max_outbound_read_limit; + props->atomic_cap = IB_ATOMIC_NONE; + props->masked_atomic_cap = IB_ATOMIC_NONE; + props->max_ah = INT_MAX; + props->max_pkeys = 1; + props->local_ca_ack_delay = MANA_CA_ACK_DELAY; return 0; } @@ -654,6 +667,33 @@ int mana_ib_gd_query_adapter_caps(struct mana_ib_dev *dev) return 0; } +static void +mana_ib_event_handler(void *ctx, struct gdma_queue *q, struct gdma_event *event) +{ + struct mana_ib_dev *mdev = (struct mana_ib_dev *)ctx; + struct mana_ib_qp *qp; + struct ib_event ev; + u32 qpn; + + switch (event->type) { + case GDMA_EQE_RNIC_QP_FATAL: + qpn = event->details[0]; + qp = mana_get_qp_ref(mdev, qpn); + if (!qp) + break; + if (qp->ibqp.event_handler) { + ev.device = qp->ibqp.device; + ev.element.qp = &qp->ibqp; + ev.event = IB_EVENT_QP_FATAL; + qp->ibqp.event_handler(&ev, qp->ibqp.qp_context); + } + mana_put_qp_ref(qp); + break; + default: + break; + } +} + int mana_ib_create_eqs(struct mana_ib_dev *mdev) { struct gdma_context *gc = mdev_to_gc(mdev); @@ -663,7 +703,7 @@ int mana_ib_create_eqs(struct mana_ib_dev *mdev) spec.type = GDMA_EQ; spec.monitor_avl_buf = false; spec.queue_size = EQ_SIZE; - spec.eq.callback = NULL; + spec.eq.callback = mana_ib_event_handler; spec.eq.context = mdev; spec.eq.log2_throttle_limit = LOG2_EQ_THROTTLE; spec.eq.msix_index = 0; @@ -678,7 +718,7 @@ int mana_ib_create_eqs(struct mana_ib_dev *mdev) err = -ENOMEM; goto destroy_fatal_eq; } - + spec.eq.callback = NULL; for (i = 0; i < mdev->ib_dev.num_comp_vectors; i++) { spec.eq.msix_index = (i + 1) % gc->num_msix_usable; err = mana_gd_create_mana_eq(mdev->gdma_dev, &spec, &mdev->eqs[i]); @@ -888,3 +928,62 @@ int mana_ib_gd_destroy_cq(struct mana_ib_dev *mdev, struct mana_ib_cq *cq) return 0; } + +int mana_ib_gd_create_rc_qp(struct mana_ib_dev *mdev, struct mana_ib_qp *qp, + struct ib_qp_init_attr *attr, u32 doorbell, u64 flags) +{ + struct mana_ib_cq *send_cq = container_of(qp->ibqp.send_cq, struct mana_ib_cq, ibcq); + struct mana_ib_cq *recv_cq = container_of(qp->ibqp.recv_cq, struct mana_ib_cq, ibcq); + struct mana_ib_pd *pd = container_of(qp->ibqp.pd, struct mana_ib_pd, ibpd); + struct gdma_context *gc = mdev_to_gc(mdev); + struct mana_rnic_create_qp_resp resp = {}; + struct mana_rnic_create_qp_req req = {}; + int err, i; + + mana_gd_init_req_hdr(&req.hdr, MANA_IB_CREATE_RC_QP, sizeof(req), sizeof(resp)); + req.hdr.dev_id = gc->mana_ib.dev_id; + req.adapter = mdev->adapter_handle; + req.pd_handle = pd->pd_handle; + req.send_cq_handle = send_cq->cq_handle; + req.recv_cq_handle = recv_cq->cq_handle; + for (i = 0; i < MANA_RC_QUEUE_TYPE_MAX; i++) + req.dma_region[i] = qp->rc_qp.queues[i].gdma_region; + req.doorbell_page = doorbell; + req.max_send_wr = attr->cap.max_send_wr; + req.max_recv_wr = attr->cap.max_recv_wr; + req.max_send_sge = attr->cap.max_send_sge; + req.max_recv_sge = attr->cap.max_recv_sge; + req.flags = flags; + + err = mana_gd_send_request(gc, sizeof(req), &req, sizeof(resp), &resp); + if (err) { + ibdev_err(&mdev->ib_dev, "Failed to create rc qp err %d", err); + return err; + } + qp->qp_handle = resp.rc_qp_handle; + for (i = 0; i < MANA_RC_QUEUE_TYPE_MAX; i++) { + qp->rc_qp.queues[i].id = resp.queue_ids[i]; + /* The GDMA regions are now owned by the RNIC QP handle */ + qp->rc_qp.queues[i].gdma_region = GDMA_INVALID_DMA_REGION; + } + return 0; +} + +int mana_ib_gd_destroy_rc_qp(struct mana_ib_dev *mdev, struct mana_ib_qp *qp) +{ + struct mana_rnic_destroy_rc_qp_resp resp = {0}; + struct mana_rnic_destroy_rc_qp_req req = {0}; + struct gdma_context *gc = mdev_to_gc(mdev); + int err; + + mana_gd_init_req_hdr(&req.hdr, MANA_IB_DESTROY_RC_QP, sizeof(req), sizeof(resp)); + req.hdr.dev_id = gc->mana_ib.dev_id; + req.adapter = mdev->adapter_handle; + req.rc_qp_handle = qp->qp_handle; + err = mana_gd_send_request(gc, sizeof(req), &req, sizeof(resp), &resp); + if (err) { + ibdev_err(&mdev->ib_dev, "Failed to destroy rc qp err %d", err); + return err; + } + return 0; +} diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h index 68c3b4f0faa4..977da9569701 100644 --- a/drivers/infiniband/hw/mana/mana_ib.h +++ b/drivers/infiniband/hw/mana/mana_ib.h @@ -27,6 +27,11 @@ */ #define MANA_IB_MAX_MR 0xFFFFFFu +/* + * The CA timeout is approx. 260ms (4us * 2^(DELAY)) + */ +#define MANA_CA_ACK_DELAY 16 + struct mana_ib_adapter_caps { u32 max_sq_id; u32 max_rq_id; @@ -57,6 +62,7 @@ struct mana_ib_dev { mana_handle_t adapter_handle; struct gdma_queue *fatal_err_eq; struct gdma_queue **eqs; + struct xarray qp_table_wq; struct mana_ib_adapter_caps adapter_caps; }; @@ -95,14 +101,33 @@ struct mana_ib_cq { mana_handle_t cq_handle; }; +enum mana_rc_queue_type { + MANA_RC_SEND_QUEUE_REQUESTER = 0, + MANA_RC_SEND_QUEUE_RESPONDER, + MANA_RC_SEND_QUEUE_FMR, + MANA_RC_RECV_QUEUE_REQUESTER, + MANA_RC_RECV_QUEUE_RESPONDER, + MANA_RC_QUEUE_TYPE_MAX, +}; + +struct mana_ib_rc_qp { + struct mana_ib_queue queues[MANA_RC_QUEUE_TYPE_MAX]; +}; + struct mana_ib_qp { struct ib_qp ibqp; mana_handle_t qp_handle; - struct mana_ib_queue raw_sq; + union { + struct mana_ib_queue raw_sq; + struct mana_ib_rc_qp rc_qp; + }; /* The port on the IB device, starting with 1 */ u32 port; + + refcount_t refcount; + struct completion free; }; struct mana_ib_ucontext { @@ -122,6 +147,9 @@ enum mana_ib_command_code { MANA_IB_CONFIG_MAC_ADDR = 0x30005, MANA_IB_CREATE_CQ = 0x30008, MANA_IB_DESTROY_CQ = 0x30009, + MANA_IB_CREATE_RC_QP = 0x3000a, + MANA_IB_DESTROY_RC_QP = 0x3000b, + MANA_IB_SET_QP_STATE = 0x3000d, }; struct mana_ib_query_adapter_caps_req { @@ -230,11 +258,105 @@ struct mana_rnic_destroy_cq_resp { struct gdma_resp_hdr hdr; }; /* HW Data */ +enum mana_rnic_create_rc_flags { + MANA_RC_FLAG_NO_FMR = 2, +}; + +struct mana_rnic_create_qp_req { + struct gdma_req_hdr hdr; + mana_handle_t adapter; + mana_handle_t pd_handle; + mana_handle_t send_cq_handle; + mana_handle_t recv_cq_handle; + u64 dma_region[MANA_RC_QUEUE_TYPE_MAX]; + u64 deprecated[2]; + u64 flags; + u32 doorbell_page; + u32 max_send_wr; + u32 max_recv_wr; + u32 max_send_sge; + u32 max_recv_sge; + u32 reserved; +}; /* HW Data */ + +struct mana_rnic_create_qp_resp { + struct gdma_resp_hdr hdr; + mana_handle_t rc_qp_handle; + u32 queue_ids[MANA_RC_QUEUE_TYPE_MAX]; + u32 reserved; +}; /* HW Data*/ + +struct mana_rnic_destroy_rc_qp_req { + struct gdma_req_hdr hdr; + mana_handle_t adapter; + mana_handle_t rc_qp_handle; +}; /* HW Data */ + +struct mana_rnic_destroy_rc_qp_resp { + struct gdma_resp_hdr hdr; +}; /* HW Data */ + +struct mana_ib_ah_attr { + u8 src_addr[16]; + u8 dest_addr[16]; + u8 src_mac[ETH_ALEN]; + u8 dest_mac[ETH_ALEN]; + u8 src_addr_type; + u8 dest_addr_type; + u8 hop_limit; + u8 traffic_class; + u16 src_port; + u16 dest_port; + u32 reserved; +}; + +struct mana_rnic_set_qp_state_req { + struct gdma_req_hdr hdr; + mana_handle_t adapter; + mana_handle_t qp_handle; + u64 attr_mask; + u32 qp_state; + u32 path_mtu; + u32 rq_psn; + u32 sq_psn; + u32 dest_qpn; + u32 max_dest_rd_atomic; + u32 retry_cnt; + u32 rnr_retry; + u32 min_rnr_timer; + u32 reserved; + struct mana_ib_ah_attr ah_attr; +}; /* HW Data */ + +struct mana_rnic_set_qp_state_resp { + struct gdma_resp_hdr hdr; +}; /* HW Data */ + static inline struct gdma_context *mdev_to_gc(struct mana_ib_dev *mdev) { return mdev->gdma_dev->gdma_context; } +static inline struct mana_ib_qp *mana_get_qp_ref(struct mana_ib_dev *mdev, + uint32_t qid) +{ + struct mana_ib_qp *qp; + unsigned long flag; + + xa_lock_irqsave(&mdev->qp_table_wq, flag); + qp = xa_load(&mdev->qp_table_wq, qid); + if (qp) + refcount_inc(&qp->refcount); + xa_unlock_irqrestore(&mdev->qp_table_wq, flag); + return qp; +} + +static inline void mana_put_qp_ref(struct mana_ib_qp *qp) +{ + if (refcount_dec_and_test(&qp->refcount)) + complete(&qp->free); +} + static inline struct net_device *mana_ib_get_netdev(struct ib_device *ibdev, u32 port) { struct mana_ib_dev *mdev = container_of(ibdev, struct mana_ib_dev, ib_dev); @@ -354,4 +476,8 @@ int mana_ib_gd_config_mac(struct mana_ib_dev *mdev, enum mana_ib_addr_op op, u8 int mana_ib_gd_create_cq(struct mana_ib_dev *mdev, struct mana_ib_cq *cq, u32 doorbell); int mana_ib_gd_destroy_cq(struct mana_ib_dev *mdev, struct mana_ib_cq *cq); + +int mana_ib_gd_create_rc_qp(struct mana_ib_dev *mdev, struct mana_ib_qp *qp, + struct ib_qp_init_attr *attr, u32 doorbell, u64 flags); +int mana_ib_gd_destroy_rc_qp(struct mana_ib_dev *mdev, struct mana_ib_qp *qp); #endif diff --git a/drivers/infiniband/hw/mana/qp.c b/drivers/infiniband/hw/mana/qp.c index ba13c5abf8ef..73d67c853b6f 100644 --- a/drivers/infiniband/hw/mana/qp.c +++ b/drivers/infiniband/hw/mana/qp.c @@ -21,7 +21,7 @@ static int mana_ib_cfg_vport_steering(struct mana_ib_dev *dev, gc = mdev_to_gc(dev); - req_buf_size = struct_size(req, indir_tab, MANA_INDIRECT_TABLE_SIZE); + req_buf_size = struct_size(req, indir_tab, MANA_INDIRECT_TABLE_DEF_SIZE); req = kzalloc(req_buf_size, GFP_KERNEL); if (!req) return -ENOMEM; @@ -41,18 +41,18 @@ static int mana_ib_cfg_vport_steering(struct mana_ib_dev *dev, if (log_ind_tbl_size) req->rss_enable = true; - req->num_indir_entries = MANA_INDIRECT_TABLE_SIZE; + req->num_indir_entries = MANA_INDIRECT_TABLE_DEF_SIZE; req->indir_tab_offset = offsetof(struct mana_cfg_rx_steer_req_v2, indir_tab); req->update_indir_tab = true; req->cqe_coalescing_enable = 1; /* The ind table passed to the hardware must have - * MANA_INDIRECT_TABLE_SIZE entries. Adjust the verb + * MANA_INDIRECT_TABLE_DEF_SIZE entries. Adjust the verb * ind_table to MANA_INDIRECT_TABLE_SIZE if required */ ibdev_dbg(&dev->ib_dev, "ind table size %u\n", 1 << log_ind_tbl_size); - for (i = 0; i < MANA_INDIRECT_TABLE_SIZE; i++) { + for (i = 0; i < MANA_INDIRECT_TABLE_DEF_SIZE; i++) { req->indir_tab[i] = ind_table[i % (1 << log_ind_tbl_size)]; ibdev_dbg(&dev->ib_dev, "index %u handle 0x%llx\n", i, req->indir_tab[i]); @@ -137,7 +137,7 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd, } ind_tbl_size = 1 << ind_tbl->log_ind_tbl_size; - if (ind_tbl_size > MANA_INDIRECT_TABLE_SIZE) { + if (ind_tbl_size > MANA_INDIRECT_TABLE_DEF_SIZE) { ibdev_dbg(&mdev->ib_dev, "Indirect table size %d exceeding limit\n", ind_tbl_size); @@ -398,6 +398,98 @@ err_free_vport: return err; } +static int mana_table_store_qp(struct mana_ib_dev *mdev, struct mana_ib_qp *qp) +{ + refcount_set(&qp->refcount, 1); + init_completion(&qp->free); + return xa_insert_irq(&mdev->qp_table_wq, qp->ibqp.qp_num, qp, + GFP_KERNEL); +} + +static void mana_table_remove_qp(struct mana_ib_dev *mdev, + struct mana_ib_qp *qp) +{ + xa_erase_irq(&mdev->qp_table_wq, qp->ibqp.qp_num); + mana_put_qp_ref(qp); + wait_for_completion(&qp->free); +} + +static int mana_ib_create_rc_qp(struct ib_qp *ibqp, struct ib_pd *ibpd, + struct ib_qp_init_attr *attr, struct ib_udata *udata) +{ + struct mana_ib_dev *mdev = container_of(ibpd->device, struct mana_ib_dev, ib_dev); + struct mana_ib_qp *qp = container_of(ibqp, struct mana_ib_qp, ibqp); + struct mana_ib_create_rc_qp_resp resp = {}; + struct mana_ib_ucontext *mana_ucontext; + struct mana_ib_create_rc_qp ucmd = {}; + int i, err, j; + u64 flags = 0; + u32 doorbell; + + if (!udata || udata->inlen < sizeof(ucmd)) + return -EINVAL; + + mana_ucontext = rdma_udata_to_drv_context(udata, struct mana_ib_ucontext, ibucontext); + doorbell = mana_ucontext->doorbell; + flags = MANA_RC_FLAG_NO_FMR; + err = ib_copy_from_udata(&ucmd, udata, min(sizeof(ucmd), udata->inlen)); + if (err) { + ibdev_dbg(&mdev->ib_dev, "Failed to copy from udata, %d\n", err); + return err; + } + + for (i = 0, j = 0; i < MANA_RC_QUEUE_TYPE_MAX; ++i) { + /* skip FMR for user-level RC QPs */ + if (i == MANA_RC_SEND_QUEUE_FMR) { + qp->rc_qp.queues[i].id = INVALID_QUEUE_ID; + qp->rc_qp.queues[i].gdma_region = GDMA_INVALID_DMA_REGION; + continue; + } + err = mana_ib_create_queue(mdev, ucmd.queue_buf[j], ucmd.queue_size[j], + &qp->rc_qp.queues[i]); + if (err) { + ibdev_err(&mdev->ib_dev, "Failed to create queue %d, err %d\n", i, err); + goto destroy_queues; + } + j++; + } + + err = mana_ib_gd_create_rc_qp(mdev, qp, attr, doorbell, flags); + if (err) { + ibdev_err(&mdev->ib_dev, "Failed to create rc qp %d\n", err); + goto destroy_queues; + } + qp->ibqp.qp_num = qp->rc_qp.queues[MANA_RC_RECV_QUEUE_RESPONDER].id; + qp->port = attr->port_num; + + if (udata) { + for (i = 0, j = 0; i < MANA_RC_QUEUE_TYPE_MAX; ++i) { + if (i == MANA_RC_SEND_QUEUE_FMR) + continue; + resp.queue_id[j] = qp->rc_qp.queues[i].id; + j++; + } + err = ib_copy_to_udata(udata, &resp, min(sizeof(resp), udata->outlen)); + if (err) { + ibdev_dbg(&mdev->ib_dev, "Failed to copy to udata, %d\n", err); + goto destroy_qp; + } + } + + err = mana_table_store_qp(mdev, qp); + if (err) + goto destroy_qp; + + return 0; + +destroy_qp: + mana_ib_gd_destroy_rc_qp(mdev, qp); +destroy_queues: + while (i-- > 0) + mana_ib_destroy_queue(mdev, &qp->rc_qp.queues[i]); + return err; +} + int mana_ib_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *attr, struct ib_udata *udata) { @@ -409,8 +501,9 @@ int mana_ib_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *attr, udata); return mana_ib_create_qp_raw(ibqp, ibqp->pd, attr, udata); + case IB_QPT_RC: + return mana_ib_create_rc_qp(ibqp, ibqp->pd, attr, udata); default: - /* Creating QP other than IB_QPT_RAW_PACKET is not supported */ ibdev_dbg(ibqp->device, "Creating QP type %u not supported\n", attr->qp_type); } @@ -418,11 +511,79 @@ int mana_ib_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *attr, return -EINVAL; } +static int mana_ib_gd_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, + int attr_mask, struct ib_udata *udata) +{ + struct mana_ib_dev *mdev = container_of(ibqp->device, struct mana_ib_dev, ib_dev); + struct mana_ib_qp *qp = container_of(ibqp, struct mana_ib_qp, ibqp); + struct mana_rnic_set_qp_state_resp resp = {}; + struct mana_rnic_set_qp_state_req req = {}; + struct gdma_context *gc = mdev_to_gc(mdev); + struct mana_port_context *mpc; + struct net_device *ndev; + int err; + + mana_gd_init_req_hdr(&req.hdr, MANA_IB_SET_QP_STATE, sizeof(req), sizeof(resp)); + req.hdr.dev_id = gc->mana_ib.dev_id; + req.adapter = mdev->adapter_handle; + req.qp_handle = qp->qp_handle; + req.qp_state = attr->qp_state; + req.attr_mask = attr_mask; + req.path_mtu = attr->path_mtu; + req.rq_psn = attr->rq_psn; + req.sq_psn = attr->sq_psn; + req.dest_qpn = attr->dest_qp_num; + req.max_dest_rd_atomic = attr->max_dest_rd_atomic; + req.retry_cnt = attr->retry_cnt; + req.rnr_retry = attr->rnr_retry; + req.min_rnr_timer = attr->min_rnr_timer; + if (attr_mask & IB_QP_AV) { + ndev = mana_ib_get_netdev(&mdev->ib_dev, ibqp->port); + if (!ndev) { + ibdev_dbg(&mdev->ib_dev, "Invalid port %u in QP %u\n", + ibqp->port, ibqp->qp_num); + return -EINVAL; + } + mpc = netdev_priv(ndev); + copy_in_reverse(req.ah_attr.src_mac, mpc->mac_addr, ETH_ALEN); + copy_in_reverse(req.ah_attr.dest_mac, attr->ah_attr.roce.dmac, ETH_ALEN); + copy_in_reverse(req.ah_attr.src_addr, attr->ah_attr.grh.sgid_attr->gid.raw, + sizeof(union ib_gid)); + copy_in_reverse(req.ah_attr.dest_addr, attr->ah_attr.grh.dgid.raw, + sizeof(union ib_gid)); + if (rdma_gid_attr_network_type(attr->ah_attr.grh.sgid_attr) == RDMA_NETWORK_IPV4) { + req.ah_attr.src_addr_type = SGID_TYPE_IPV4; + req.ah_attr.dest_addr_type = SGID_TYPE_IPV4; + } else { + req.ah_attr.src_addr_type = SGID_TYPE_IPV6; + req.ah_attr.dest_addr_type = SGID_TYPE_IPV6; + } + req.ah_attr.dest_port = ROCE_V2_UDP_DPORT; + req.ah_attr.src_port = rdma_get_udp_sport(attr->ah_attr.grh.flow_label, + ibqp->qp_num, attr->dest_qp_num); + req.ah_attr.traffic_class = attr->ah_attr.grh.traffic_class; + req.ah_attr.hop_limit = attr->ah_attr.grh.hop_limit; + } + + err = mana_gd_send_request(gc, sizeof(req), &req, sizeof(resp), &resp); + if (err) { + ibdev_err(&mdev->ib_dev, "Failed modify qp err %d", err); + return err; + } + + return 0; +} + int mana_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, int attr_mask, struct ib_udata *udata) { - /* modify_qp is not supported by this version of the driver */ - return -EOPNOTSUPP; + switch (ibqp->qp_type) { + case IB_QPT_RC: + return mana_ib_gd_modify_qp(ibqp, attr, attr_mask, udata); + default: + ibdev_dbg(ibqp->device, "Modify QP type %u not supported", ibqp->qp_type); + return -EOPNOTSUPP; + } } static int mana_ib_destroy_qp_rss(struct mana_ib_qp *qp, @@ -473,6 +634,24 @@ static int mana_ib_destroy_qp_raw(struct mana_ib_qp *qp, struct ib_udata *udata) return 0; } +static int mana_ib_destroy_rc_qp(struct mana_ib_qp *qp, struct ib_udata *udata) +{ + struct mana_ib_dev *mdev = + container_of(qp->ibqp.device, struct mana_ib_dev, ib_dev); + int i; + + mana_table_remove_qp(mdev, qp); + + /* Ignore return code as there is not much we can do about it. + * The error message is printed inside. + */ + mana_ib_gd_destroy_rc_qp(mdev, qp); + for (i = 0; i < MANA_RC_QUEUE_TYPE_MAX; ++i) + mana_ib_destroy_queue(mdev, &qp->rc_qp.queues[i]); + + return 0; +} + int mana_ib_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata) { struct mana_ib_qp *qp = container_of(ibqp, struct mana_ib_qp, ibqp); @@ -484,7 +663,8 @@ int mana_ib_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata) udata); return mana_ib_destroy_qp_raw(qp, udata); - + case IB_QPT_RC: + return mana_ib_destroy_rc_qp(qp, udata); default: ibdev_dbg(ibqp->device, "Unexpected QP type %u\n", ibqp->qp_type); diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index 00e292422801..cee167cf310e 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -2823,37 +2823,72 @@ static u8 mlx5_get_umr_fence(u8 umr_fence_cap) } } -static int mlx5_ib_dev_res_init(struct mlx5_ib_dev *dev) +int mlx5_ib_dev_res_cq_init(struct mlx5_ib_dev *dev) { struct mlx5_ib_resources *devr = &dev->devr; - struct ib_srq_init_attr attr; - struct ib_device *ibdev; struct ib_cq_init_attr cq_attr = {.cqe = 1}; - int port; + struct ib_device *ibdev; + struct ib_pd *pd; + struct ib_cq *cq; int ret = 0; - ibdev = &dev->ib_dev; - if (!MLX5_CAP_GEN(dev->mdev, xrc)) - return -EOPNOTSUPP; + /* + * devr->c0 is set once, never changed until device unload. + * Avoid taking the mutex if initialization is already done. + */ + if (devr->c0) + return 0; - devr->p0 = ib_alloc_pd(ibdev, 0); - if (IS_ERR(devr->p0)) - return PTR_ERR(devr->p0); + mutex_lock(&devr->cq_lock); + if (devr->c0) + goto unlock; - devr->c0 = ib_create_cq(ibdev, NULL, NULL, NULL, &cq_attr); - if (IS_ERR(devr->c0)) { - ret = PTR_ERR(devr->c0); - goto error1; + ibdev = &dev->ib_dev; + pd = ib_alloc_pd(ibdev, 0); + if (IS_ERR(pd)) { + ret = PTR_ERR(pd); + mlx5_ib_err(dev, "Couldn't allocate PD for res init, err=%d\n", ret); + goto unlock; } - ret = mlx5_cmd_xrcd_alloc(dev->mdev, &devr->xrcdn0, 0); - if (ret) - goto error2; + cq = ib_create_cq(ibdev, NULL, NULL, NULL, &cq_attr); + if (IS_ERR(cq)) { + ret = PTR_ERR(cq); + mlx5_ib_err(dev, "Couldn't create CQ for res init, err=%d\n", ret); + ib_dealloc_pd(pd); + goto unlock; + } - ret = mlx5_cmd_xrcd_alloc(dev->mdev, &devr->xrcdn1, 0); + devr->p0 = pd; + devr->c0 = cq; + +unlock: + mutex_unlock(&devr->cq_lock); + return ret; +} + +int mlx5_ib_dev_res_srq_init(struct mlx5_ib_dev *dev) +{ + struct mlx5_ib_resources *devr = &dev->devr; + struct ib_srq_init_attr attr; + struct ib_srq *s0, *s1; + int ret = 0; + + /* + * devr->s1 is set once, never changed until device unload. + * Avoid taking the mutex if initialization is already done. + */ + if (devr->s1) + return 0; + + mutex_lock(&devr->srq_lock); + if (devr->s1) + goto unlock; + + ret = mlx5_ib_dev_res_cq_init(dev); if (ret) - goto error3; + goto unlock; memset(&attr, 0, sizeof(attr)); attr.attr.max_sge = 1; @@ -2861,10 +2896,11 @@ static int mlx5_ib_dev_res_init(struct mlx5_ib_dev *dev) attr.srq_type = IB_SRQT_XRC; attr.ext.cq = devr->c0; - devr->s0 = ib_create_srq(devr->p0, &attr); - if (IS_ERR(devr->s0)) { - ret = PTR_ERR(devr->s0); - goto err_create; + s0 = ib_create_srq(devr->p0, &attr); + if (IS_ERR(s0)) { + ret = PTR_ERR(s0); + mlx5_ib_err(dev, "Couldn't create SRQ 0 for res init, err=%d\n", ret); + goto unlock; } memset(&attr, 0, sizeof(attr)); @@ -2872,29 +2908,48 @@ static int mlx5_ib_dev_res_init(struct mlx5_ib_dev *dev) attr.attr.max_wr = 1; attr.srq_type = IB_SRQT_BASIC; - devr->s1 = ib_create_srq(devr->p0, &attr); - if (IS_ERR(devr->s1)) { - ret = PTR_ERR(devr->s1); - goto error6; + s1 = ib_create_srq(devr->p0, &attr); + if (IS_ERR(s1)) { + ret = PTR_ERR(s1); + mlx5_ib_err(dev, "Couldn't create SRQ 1 for res init, err=%d\n", ret); + ib_destroy_srq(s0); + } + + devr->s0 = s0; + devr->s1 = s1; + +unlock: + mutex_unlock(&devr->srq_lock); + return ret; +} + +static int mlx5_ib_dev_res_init(struct mlx5_ib_dev *dev) +{ + struct mlx5_ib_resources *devr = &dev->devr; + int port; + int ret; + + if (!MLX5_CAP_GEN(dev->mdev, xrc)) + return -EOPNOTSUPP; + + ret = mlx5_cmd_xrcd_alloc(dev->mdev, &devr->xrcdn0, 0); + if (ret) + return ret; + + ret = mlx5_cmd_xrcd_alloc(dev->mdev, &devr->xrcdn1, 0); + if (ret) { + mlx5_cmd_xrcd_dealloc(dev->mdev, devr->xrcdn0, 0); + return ret; } for (port = 0; port < ARRAY_SIZE(devr->ports); ++port) INIT_WORK(&devr->ports[port].pkey_change_work, pkey_change_handler); - return 0; + mutex_init(&devr->cq_lock); + mutex_init(&devr->srq_lock); -error6: - ib_destroy_srq(devr->s0); -err_create: - mlx5_cmd_xrcd_dealloc(dev->mdev, devr->xrcdn1, 0); -error3: - mlx5_cmd_xrcd_dealloc(dev->mdev, devr->xrcdn0, 0); -error2: - ib_destroy_cq(devr->c0); -error1: - ib_dealloc_pd(devr->p0); - return ret; + return 0; } static void mlx5_ib_dev_res_cleanup(struct mlx5_ib_dev *dev) @@ -2911,12 +2966,20 @@ static void mlx5_ib_dev_res_cleanup(struct mlx5_ib_dev *dev) for (port = 0; port < ARRAY_SIZE(devr->ports); ++port) cancel_work_sync(&devr->ports[port].pkey_change_work); - ib_destroy_srq(devr->s1); - ib_destroy_srq(devr->s0); + /* After s0/s1 init, they are not unset during the device lifetime. */ + if (devr->s1) { + ib_destroy_srq(devr->s1); + ib_destroy_srq(devr->s0); + } mlx5_cmd_xrcd_dealloc(dev->mdev, devr->xrcdn1, 0); mlx5_cmd_xrcd_dealloc(dev->mdev, devr->xrcdn0, 0); - ib_destroy_cq(devr->c0); - ib_dealloc_pd(devr->p0); + /* After p0/c0 init, they are not unset during the device lifetime. */ + if (devr->c0) { + ib_destroy_cq(devr->c0); + ib_dealloc_pd(devr->p0); + } + mutex_destroy(&devr->cq_lock); + mutex_destroy(&devr->srq_lock); } static u32 get_core_cap_flags(struct ib_device *ibdev, @@ -4086,6 +4149,7 @@ static void mlx5_ib_stage_pre_ib_reg_umr_cleanup(struct mlx5_ib_dev *dev) { mlx5_mkey_cache_cleanup(dev); mlx5r_umr_resource_cleanup(dev); + mlx5r_umr_cleanup(dev); } static void mlx5_ib_stage_ib_reg_cleanup(struct mlx5_ib_dev *dev) @@ -4097,7 +4161,7 @@ static int mlx5_ib_stage_post_ib_reg_umr_init(struct mlx5_ib_dev *dev) { int ret; - ret = mlx5r_umr_resource_init(dev); + ret = mlx5r_umr_init(dev); if (ret) return ret; diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h index b68779e9d86c..74a20df8b292 100644 --- a/drivers/infiniband/hw/mlx5/mlx5_ib.h +++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h @@ -115,6 +115,19 @@ unsigned long __mlx5_umem_find_best_quantized_pgoff( __mlx5_bit_sz(typ, page_offset_fld), 0, scale, \ page_offset_quantized) +static inline unsigned long +mlx5_umem_dmabuf_find_best_pgsz(struct ib_umem_dmabuf *umem_dmabuf) +{ + /* + * mkeys used for dmabuf are fixed at PAGE_SIZE because we must be able + * to hold any sgl after a move operation. Ideally the mkc page size + * could be changed at runtime to be optimal, but right now the driver + * cannot do that. + */ + return ib_umem_find_best_pgsz(&umem_dmabuf->umem, PAGE_SIZE, + umem_dmabuf->umem.iova); +} + enum { MLX5_IB_MMAP_OFFSET_START = 9, MLX5_IB_MMAP_OFFSET_END = 255, @@ -751,6 +764,8 @@ struct umr_common { */ struct mutex lock; unsigned int state; + /* Protects from repeat UMR QP creation */ + struct mutex init_lock; }; #define NUM_MKEYS_PER_PAGE \ @@ -822,11 +837,13 @@ struct mlx5_ib_port_resources { struct mlx5_ib_resources { struct ib_cq *c0; + struct mutex cq_lock; u32 xrcdn0; u32 xrcdn1; struct ib_pd *p0; struct ib_srq *s0; struct ib_srq *s1; + struct mutex srq_lock; struct mlx5_ib_port_resources ports[2]; }; @@ -1270,6 +1287,8 @@ to_mmmap(struct rdma_user_mmap_entry *rdma_entry) struct mlx5_user_mmap_entry, rdma_entry); } +int mlx5_ib_dev_res_cq_init(struct mlx5_ib_dev *dev); +int mlx5_ib_dev_res_srq_init(struct mlx5_ib_dev *dev); int mlx5_ib_db_map_user(struct mlx5_ib_ucontext *context, unsigned long virt, struct mlx5_db *db); void mlx5_ib_db_unmap_user(struct mlx5_ib_ucontext *context, struct mlx5_db *db); diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c index ecc111ed5d86..bc9c3871f715 100644 --- a/drivers/infiniband/hw/mlx5/mr.c +++ b/drivers/infiniband/hw/mlx5/mr.c @@ -1470,6 +1470,7 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, { struct mlx5_ib_dev *dev = to_mdev(pd->device); struct ib_umem *umem; + int err; if (!IS_ENABLED(CONFIG_INFINIBAND_USER_MEM)) return ERR_PTR(-EOPNOTSUPP); @@ -1477,6 +1478,10 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, mlx5_ib_dbg(dev, "start 0x%llx, iova 0x%llx, length 0x%llx, access_flags 0x%x\n", start, iova, length, access_flags); + err = mlx5r_umr_resource_init(dev); + if (err) + return ERR_PTR(err); + if (access_flags & IB_ACCESS_ON_DEMAND) return create_user_odp_mr(pd, start, length, iova, access_flags, udata); @@ -1523,6 +1528,10 @@ struct ib_mr *mlx5_ib_reg_user_mr_dmabuf(struct ib_pd *pd, u64 offset, "offset 0x%llx, virt_addr 0x%llx, length 0x%llx, fd %d, access_flags 0x%x\n", offset, virt_addr, length, fd, access_flags); + err = mlx5r_umr_resource_init(dev); + if (err) + return ERR_PTR(err); + /* dmabuf requires xlt update via umr to work. */ if (!mlx5r_umr_can_load_pas(dev, length)) return ERR_PTR(-EINVAL); diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c index 4a04cbc5b78a..a524181f34df 100644 --- a/drivers/infiniband/hw/mlx5/odp.c +++ b/drivers/infiniband/hw/mlx5/odp.c @@ -705,10 +705,8 @@ static int pagefault_dmabuf_mr(struct mlx5_ib_mr *mr, size_t bcnt, return err; } - page_size = mlx5_umem_find_best_pgsz(&umem_dmabuf->umem, mkc, - log_page_size, 0, - umem_dmabuf->umem.iova); - if (unlikely(page_size < PAGE_SIZE)) { + page_size = mlx5_umem_dmabuf_find_best_pgsz(umem_dmabuf); + if (!page_size) { ib_umem_dmabuf_unmap_pages(umem_dmabuf); err = -EINVAL; } else { diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index e8c0fead4062..079ad927182f 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -3234,6 +3234,10 @@ int mlx5_ib_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *attr, enum ib_qp_type type; int err; + err = mlx5_ib_dev_res_srq_init(dev); + if (err) + return err; + err = check_qp_type(dev, attr, &type); if (err) return err; diff --git a/drivers/infiniband/hw/mlx5/srq.c b/drivers/infiniband/hw/mlx5/srq.c index a056ea835da5..fe907c908d82 100644 --- a/drivers/infiniband/hw/mlx5/srq.c +++ b/drivers/infiniband/hw/mlx5/srq.c @@ -213,6 +213,10 @@ int mlx5_ib_create_srq(struct ib_srq *ib_srq, return -EINVAL; } + err = mlx5_ib_dev_res_cq_init(dev); + if (err) + return err; + mutex_init(&srq->mutex); spin_lock_init(&srq->lock); srq->msrq.max = roundup_pow_of_two(init_attr->attr.max_wr + 1); diff --git a/drivers/infiniband/hw/mlx5/umr.c b/drivers/infiniband/hw/mlx5/umr.c index e76142f6fa88..ffc31b01f690 100644 --- a/drivers/infiniband/hw/mlx5/umr.c +++ b/drivers/infiniband/hw/mlx5/umr.c @@ -135,22 +135,28 @@ static int mlx5r_umr_qp_rst2rts(struct mlx5_ib_dev *dev, struct ib_qp *qp) int mlx5r_umr_resource_init(struct mlx5_ib_dev *dev) { struct ib_qp_init_attr init_attr = {}; - struct ib_pd *pd; struct ib_cq *cq; struct ib_qp *qp; - int ret; + int ret = 0; - pd = ib_alloc_pd(&dev->ib_dev, 0); - if (IS_ERR(pd)) { - mlx5_ib_dbg(dev, "Couldn't create PD for sync UMR QP\n"); - return PTR_ERR(pd); - } + + /* + * UMR qp is set once, never changed until device unload. + * Avoid taking the mutex if initialization is already done. + */ + if (dev->umrc.qp) + return 0; + + mutex_lock(&dev->umrc.init_lock); + /* First user allocates the UMR resources. Skip if already allocated. */ + if (dev->umrc.qp) + goto unlock; cq = ib_alloc_cq(&dev->ib_dev, NULL, 128, 0, IB_POLL_SOFTIRQ); if (IS_ERR(cq)) { mlx5_ib_dbg(dev, "Couldn't create CQ for sync UMR QP\n"); ret = PTR_ERR(cq); - goto destroy_pd; + goto unlock; } init_attr.send_cq = cq; @@ -160,7 +166,7 @@ int mlx5r_umr_resource_init(struct mlx5_ib_dev *dev) init_attr.cap.max_send_sge = 1; init_attr.qp_type = MLX5_IB_QPT_REG_UMR; init_attr.port_num = 1; - qp = ib_create_qp(pd, &init_attr); + qp = ib_create_qp(dev->umrc.pd, &init_attr); if (IS_ERR(qp)) { mlx5_ib_dbg(dev, "Couldn't create sync UMR QP\n"); ret = PTR_ERR(qp); @@ -171,22 +177,22 @@ int mlx5r_umr_resource_init(struct mlx5_ib_dev *dev) if (ret) goto destroy_qp; - dev->umrc.qp = qp; dev->umrc.cq = cq; - dev->umrc.pd = pd; sema_init(&dev->umrc.sem, MAX_UMR_WR); mutex_init(&dev->umrc.lock); dev->umrc.state = MLX5_UMR_STATE_ACTIVE; + dev->umrc.qp = qp; + mutex_unlock(&dev->umrc.init_lock); return 0; destroy_qp: ib_destroy_qp(qp); destroy_cq: ib_free_cq(cq); -destroy_pd: - ib_dealloc_pd(pd); +unlock: + mutex_unlock(&dev->umrc.init_lock); return ret; } @@ -194,8 +200,31 @@ void mlx5r_umr_resource_cleanup(struct mlx5_ib_dev *dev) { if (dev->umrc.state == MLX5_UMR_STATE_UNINIT) return; + mutex_destroy(&dev->umrc.lock); + /* After device init, UMR cp/qp are not unset during the lifetime. */ ib_destroy_qp(dev->umrc.qp); ib_free_cq(dev->umrc.cq); +} + +int mlx5r_umr_init(struct mlx5_ib_dev *dev) +{ + struct ib_pd *pd; + + pd = ib_alloc_pd(&dev->ib_dev, 0); + if (IS_ERR(pd)) { + mlx5_ib_dbg(dev, "Couldn't create PD for sync UMR QP\n"); + return PTR_ERR(pd); + } + dev->umrc.pd = pd; + + mutex_init(&dev->umrc.init_lock); + + return 0; +} + +void mlx5r_umr_cleanup(struct mlx5_ib_dev *dev) +{ + mutex_destroy(&dev->umrc.init_lock); ib_dealloc_pd(dev->umrc.pd); } diff --git a/drivers/infiniband/hw/mlx5/umr.h b/drivers/infiniband/hw/mlx5/umr.h index 3799bb758e49..5f734dc72bef 100644 --- a/drivers/infiniband/hw/mlx5/umr.h +++ b/drivers/infiniband/hw/mlx5/umr.h @@ -16,6 +16,9 @@ int mlx5r_umr_resource_init(struct mlx5_ib_dev *dev); void mlx5r_umr_resource_cleanup(struct mlx5_ib_dev *dev); +int mlx5r_umr_init(struct mlx5_ib_dev *dev); +void mlx5r_umr_cleanup(struct mlx5_ib_dev *dev); + static inline bool mlx5r_umr_can_load_pas(struct mlx5_ib_dev *dev, size_t length) { diff --git a/drivers/net/ethernet/microsoft/mana/gdma_main.c b/drivers/net/ethernet/microsoft/mana/gdma_main.c index 1332db9a08eb..a489297f26a0 100644 --- a/drivers/net/ethernet/microsoft/mana/gdma_main.c +++ b/drivers/net/ethernet/microsoft/mana/gdma_main.c @@ -380,6 +380,7 @@ static void mana_gd_process_eqe(struct gdma_queue *eq) case GDMA_EQE_HWC_INIT_EQ_ID_DB: case GDMA_EQE_HWC_INIT_DATA: case GDMA_EQE_HWC_INIT_DONE: + case GDMA_EQE_RNIC_QP_FATAL: if (!eq->eq.callback) break; diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c index d087cf954f75..b89ad4afd66e 100644 --- a/drivers/net/ethernet/microsoft/mana/mana_en.c +++ b/drivers/net/ethernet/microsoft/mana/mana_en.c @@ -481,7 +481,7 @@ static int mana_get_tx_queue(struct net_device *ndev, struct sk_buff *skb, struct sock *sk = skb->sk; int txq; - txq = apc->indir_table[hash & MANA_INDIRECT_TABLE_MASK]; + txq = apc->indir_table[hash & (apc->indir_table_sz - 1)]; if (txq != old_q && sk && sk_fullsock(sk) && rcu_access_pointer(sk->sk_dst_cache)) @@ -721,6 +721,13 @@ static void mana_cleanup_port_context(struct mana_port_context *apc) apc->rxqs = NULL; } +static void mana_cleanup_indir_table(struct mana_port_context *apc) +{ + apc->indir_table_sz = 0; + kfree(apc->indir_table); + kfree(apc->rxobj_table); +} + static int mana_init_port_context(struct mana_port_context *apc) { apc->rxqs = kcalloc(apc->num_queues, sizeof(struct mana_rxq *), @@ -962,7 +969,16 @@ static int mana_query_vport_cfg(struct mana_port_context *apc, u32 vport_index, *max_sq = resp.max_num_sq; *max_rq = resp.max_num_rq; - *num_indir_entry = resp.num_indirection_ent; + if (resp.num_indirection_ent > 0 && + resp.num_indirection_ent <= MANA_INDIRECT_TABLE_MAX_SIZE && + is_power_of_2(resp.num_indirection_ent)) { + *num_indir_entry = resp.num_indirection_ent; + } else { + netdev_warn(apc->ndev, + "Setting indirection table size to default %d for vPort %d\n", + MANA_INDIRECT_TABLE_DEF_SIZE, apc->port_idx); + *num_indir_entry = MANA_INDIRECT_TABLE_DEF_SIZE; + } apc->port_handle = resp.vport; ether_addr_copy(apc->mac_addr, resp.mac_addr); @@ -1054,14 +1070,13 @@ static int mana_cfg_vport_steering(struct mana_port_context *apc, bool update_default_rxobj, bool update_key, bool update_tab) { - u16 num_entries = MANA_INDIRECT_TABLE_SIZE; struct mana_cfg_rx_steer_req_v2 *req; struct mana_cfg_rx_steer_resp resp = {}; struct net_device *ndev = apc->ndev; u32 req_buf_size; int err; - req_buf_size = struct_size(req, indir_tab, num_entries); + req_buf_size = struct_size(req, indir_tab, apc->indir_table_sz); req = kzalloc(req_buf_size, GFP_KERNEL); if (!req) return -ENOMEM; @@ -1072,7 +1087,7 @@ static int mana_cfg_vport_steering(struct mana_port_context *apc, req->hdr.req.msg_version = GDMA_MESSAGE_V2; req->vport = apc->port_handle; - req->num_indir_entries = num_entries; + req->num_indir_entries = apc->indir_table_sz; req->indir_tab_offset = offsetof(struct mana_cfg_rx_steer_req_v2, indir_tab); req->rx_enable = rx; @@ -1111,7 +1126,7 @@ static int mana_cfg_vport_steering(struct mana_port_context *apc, } netdev_info(ndev, "Configured steering vPort %llu entries %u\n", - apc->port_handle, num_entries); + apc->port_handle, apc->indir_table_sz); out: kfree(req); return err; @@ -2344,11 +2359,33 @@ static int mana_create_vport(struct mana_port_context *apc, return mana_create_txq(apc, net); } +static int mana_rss_table_alloc(struct mana_port_context *apc) +{ + if (!apc->indir_table_sz) { + netdev_err(apc->ndev, + "Indirection table size not set for vPort %d\n", + apc->port_idx); + return -EINVAL; + } + + apc->indir_table = kcalloc(apc->indir_table_sz, sizeof(u32), GFP_KERNEL); + if (!apc->indir_table) + return -ENOMEM; + + apc->rxobj_table = kcalloc(apc->indir_table_sz, sizeof(mana_handle_t), GFP_KERNEL); + if (!apc->rxobj_table) { + kfree(apc->indir_table); + return -ENOMEM; + } + + return 0; +} + static void mana_rss_table_init(struct mana_port_context *apc) { int i; - for (i = 0; i < MANA_INDIRECT_TABLE_SIZE; i++) + for (i = 0; i < apc->indir_table_sz; i++) apc->indir_table[i] = ethtool_rxfh_indir_default(i, apc->num_queues); } @@ -2361,7 +2398,7 @@ int mana_config_rss(struct mana_port_context *apc, enum TRI_STATE rx, int i; if (update_tab) { - for (i = 0; i < MANA_INDIRECT_TABLE_SIZE; i++) { + for (i = 0; i < apc->indir_table_sz; i++) { queue_idx = apc->indir_table[i]; apc->rxobj_table[i] = apc->rxqs[queue_idx]->rxobj; } @@ -2466,7 +2503,6 @@ static int mana_init_port(struct net_device *ndev) struct mana_port_context *apc = netdev_priv(ndev); u32 max_txq, max_rxq, max_queues; int port_idx = apc->port_idx; - u32 num_indirect_entries; int err; err = mana_init_port_context(apc); @@ -2474,7 +2510,7 @@ static int mana_init_port(struct net_device *ndev) return err; err = mana_query_vport_cfg(apc, port_idx, &max_txq, &max_rxq, - &num_indirect_entries); + &apc->indir_table_sz); if (err) { netdev_err(ndev, "Failed to query info for vPort %d\n", port_idx); @@ -2723,6 +2759,10 @@ static int mana_probe_port(struct mana_context *ac, int port_idx, if (err) goto free_net; + err = mana_rss_table_alloc(apc); + if (err) + goto reset_apc; + netdev_lockdep_set_classes(ndev); ndev->hw_features = NETIF_F_SG | NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM; @@ -2739,11 +2779,13 @@ static int mana_probe_port(struct mana_context *ac, int port_idx, err = register_netdev(ndev); if (err) { netdev_err(ndev, "Unable to register netdev.\n"); - goto reset_apc; + goto free_indir; } return 0; +free_indir: + mana_cleanup_indir_table(apc); reset_apc: kfree(apc->rxqs); apc->rxqs = NULL; @@ -2872,16 +2914,30 @@ int mana_probe(struct gdma_dev *gd, bool resuming) if (!resuming) { for (i = 0; i < ac->num_ports; i++) { err = mana_probe_port(ac, i, &ac->ports[i]); - if (err) + /* we log the port for which the probe failed and stop + * probes for subsequent ports. + * Note that we keep running ports, for which the probes + * were successful, unless add_adev fails too + */ + if (err) { + dev_err(dev, "Probe Failed for port %d\n", i); break; + } } } else { for (i = 0; i < ac->num_ports; i++) { rtnl_lock(); err = mana_attach(ac->ports[i]); rtnl_unlock(); - if (err) + /* we log the port for which the attach failed and stop + * attach for subsequent ports + * Note that we keep running ports, for which the attach + * were successful, unless add_adev fails too + */ + if (err) { + dev_err(dev, "Attach Failed for port %d\n", i); break; + } } } @@ -2897,6 +2953,7 @@ void mana_remove(struct gdma_dev *gd, bool suspending) { struct gdma_context *gc = gd->gdma_context; struct mana_context *ac = gd->driver_data; + struct mana_port_context *apc; struct device *dev = gc->dev; struct net_device *ndev; int err; @@ -2908,6 +2965,7 @@ void mana_remove(struct gdma_dev *gd, bool suspending) for (i = 0; i < ac->num_ports; i++) { ndev = ac->ports[i]; + apc = netdev_priv(ndev); if (!ndev) { if (i == 0) dev_err(dev, "No net device to remove\n"); @@ -2931,6 +2989,7 @@ void mana_remove(struct gdma_dev *gd, bool suspending) } unregister_netdevice(ndev); + mana_cleanup_indir_table(apc); rtnl_unlock(); diff --git a/drivers/net/ethernet/microsoft/mana/mana_ethtool.c b/drivers/net/ethernet/microsoft/mana/mana_ethtool.c index ab2413d71f6c..146d5db1792f 100644 --- a/drivers/net/ethernet/microsoft/mana/mana_ethtool.c +++ b/drivers/net/ethernet/microsoft/mana/mana_ethtool.c @@ -245,7 +245,9 @@ static u32 mana_get_rxfh_key_size(struct net_device *ndev) static u32 mana_rss_indir_size(struct net_device *ndev) { - return MANA_INDIRECT_TABLE_SIZE; + struct mana_port_context *apc = netdev_priv(ndev); + + return apc->indir_table_sz; } static int mana_get_rxfh(struct net_device *ndev, @@ -257,7 +259,7 @@ static int mana_get_rxfh(struct net_device *ndev, rxfh->hfunc = ETH_RSS_HASH_TOP; /* Toeplitz */ if (rxfh->indir) { - for (i = 0; i < MANA_INDIRECT_TABLE_SIZE; i++) + for (i = 0; i < apc->indir_table_sz; i++) rxfh->indir[i] = apc->indir_table[i]; } @@ -273,8 +275,8 @@ static int mana_set_rxfh(struct net_device *ndev, { struct mana_port_context *apc = netdev_priv(ndev); bool update_hash = false, update_table = false; - u32 save_table[MANA_INDIRECT_TABLE_SIZE]; u8 save_key[MANA_HASH_KEY_SIZE]; + u32 *save_table; int i, err; if (!apc->port_is_up) @@ -284,13 +286,19 @@ static int mana_set_rxfh(struct net_device *ndev, rxfh->hfunc != ETH_RSS_HASH_TOP) return -EOPNOTSUPP; + save_table = kcalloc(apc->indir_table_sz, sizeof(u32), GFP_KERNEL); + if (!save_table) + return -ENOMEM; + if (rxfh->indir) { - for (i = 0; i < MANA_INDIRECT_TABLE_SIZE; i++) - if (rxfh->indir[i] >= apc->num_queues) - return -EINVAL; + for (i = 0; i < apc->indir_table_sz; i++) + if (rxfh->indir[i] >= apc->num_queues) { + err = -EINVAL; + goto cleanup; + } update_table = true; - for (i = 0; i < MANA_INDIRECT_TABLE_SIZE; i++) { + for (i = 0; i < apc->indir_table_sz; i++) { save_table[i] = apc->indir_table[i]; apc->indir_table[i] = rxfh->indir[i]; } @@ -306,7 +314,7 @@ static int mana_set_rxfh(struct net_device *ndev, if (err) { /* recover to original values */ if (update_table) { - for (i = 0; i < MANA_INDIRECT_TABLE_SIZE; i++) + for (i = 0; i < apc->indir_table_sz; i++) apc->indir_table[i] = save_table[i]; } @@ -316,6 +324,9 @@ static int mana_set_rxfh(struct net_device *ndev, mana_config_rss(apc, TRI_STATE_TRUE, update_hash, update_table); } +cleanup: + kfree(save_table); + return err; } diff --git a/include/net/mana/gdma.h b/include/net/mana/gdma.h index 27684135bb4d..5832081bc8dc 100644 --- a/include/net/mana/gdma.h +++ b/include/net/mana/gdma.h @@ -60,6 +60,7 @@ enum gdma_eqe_type { GDMA_EQE_HWC_INIT_DONE = 131, GDMA_EQE_HWC_SOC_RECONFIG = 132, GDMA_EQE_HWC_SOC_RECONFIG_DATA = 133, + GDMA_EQE_RNIC_QP_FATAL = 176, }; enum { @@ -543,11 +544,13 @@ enum { */ #define GDMA_DRV_CAP_FLAG_1_NAPI_WKDONE_FIX BIT(2) #define GDMA_DRV_CAP_FLAG_1_HWC_TIMEOUT_RECONFIG BIT(3) +#define GDMA_DRV_CAP_FLAG_1_VARIABLE_INDIRECTION_TABLE_SUPPORT BIT(5) #define GDMA_DRV_CAP_FLAGS1 \ (GDMA_DRV_CAP_FLAG_1_EQ_SHARING_MULTI_VPORT | \ GDMA_DRV_CAP_FLAG_1_NAPI_WKDONE_FIX | \ - GDMA_DRV_CAP_FLAG_1_HWC_TIMEOUT_RECONFIG) + GDMA_DRV_CAP_FLAG_1_HWC_TIMEOUT_RECONFIG | \ + GDMA_DRV_CAP_FLAG_1_VARIABLE_INDIRECTION_TABLE_SUPPORT) #define GDMA_DRV_CAP_FLAGS2 0 diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h index 561f6719fb4e..59823901b74f 100644 --- a/include/net/mana/mana.h +++ b/include/net/mana/mana.h @@ -30,8 +30,8 @@ enum TRI_STATE { }; /* Number of entries for hardware indirection table must be in power of 2 */ -#define MANA_INDIRECT_TABLE_SIZE 64 -#define MANA_INDIRECT_TABLE_MASK (MANA_INDIRECT_TABLE_SIZE - 1) +#define MANA_INDIRECT_TABLE_MAX_SIZE 512 +#define MANA_INDIRECT_TABLE_DEF_SIZE 64 /* The Toeplitz hash key's length in bytes: should be multiple of 8 */ #define MANA_HASH_KEY_SIZE 40 @@ -410,10 +410,11 @@ struct mana_port_context { struct mana_tx_qp *tx_qp; /* Indirection Table for RX & TX. The values are queue indexes */ - u32 indir_table[MANA_INDIRECT_TABLE_SIZE]; + u32 *indir_table; + u32 indir_table_sz; /* Indirection table containing RxObject Handles */ - mana_handle_t rxobj_table[MANA_INDIRECT_TABLE_SIZE]; + mana_handle_t *rxobj_table; /* Hash key used by the NIC */ u8 hashkey[MANA_HASH_KEY_SIZE]; diff --git a/include/uapi/rdma/bnxt_re-abi.h b/include/uapi/rdma/bnxt_re-abi.h index c0c34aca90ec..e61104f35d73 100644 --- a/include/uapi/rdma/bnxt_re-abi.h +++ b/include/uapi/rdma/bnxt_re-abi.h @@ -55,7 +55,7 @@ enum { BNXT_RE_UCNTX_CMASK_WC_DPI_ENABLED = 0x04ULL, BNXT_RE_UCNTX_CMASK_DBR_PACING_ENABLED = 0x08ULL, BNXT_RE_UCNTX_CMASK_POW2_DISABLED = 0x10ULL, - BNXT_RE_COMP_MASK_UCNTX_HW_RETX_ENABLED = 0x40, + BNXT_RE_UCNTX_CMASK_MSN_TABLE_ENABLED = 0x40, }; enum bnxt_re_wqe_mode { diff --git a/include/uapi/rdma/mana-abi.h b/include/uapi/rdma/mana-abi.h index 2c41cc315218..45c2df619f07 100644 --- a/include/uapi/rdma/mana-abi.h +++ b/include/uapi/rdma/mana-abi.h @@ -45,6 +45,15 @@ struct mana_ib_create_qp_resp { __u32 reserved; }; +struct mana_ib_create_rc_qp { + __aligned_u64 queue_buf[4]; + __u32 queue_size[4]; +}; + +struct mana_ib_create_rc_qp_resp { + __u32 queue_id[4]; +}; + struct mana_ib_create_wq { __aligned_u64 wq_buf_addr; __u32 wq_buf_size; |