From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from dpdk.org (dpdk.org [92.243.14.124]) by inbox.dpdk.org (Postfix) with ESMTP id 76500A04BC for ; Sat, 10 Oct 2020 06:07:53 +0200 (CEST) Received: from [92.243.14.124] (localhost [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id 1A09B1D6C7; Sat, 10 Oct 2020 06:07:42 +0200 (CEST) Received: from mail-ot1-f98.google.com (mail-ot1-f98.google.com [209.85.210.98]) by dpdk.org (Postfix) with ESMTP id 637341D683 for ; Sat, 10 Oct 2020 06:07:35 +0200 (CEST) Received: by mail-ot1-f98.google.com with SMTP id m13so10840675otl.9 for ; Fri, 09 Oct 2020 21:07:35 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=broadcom.com; s=google; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=H49uHC+ovBJNcWDLUzyV16ecJUPkdKoVHPVznFUY7Cs=; b=USuYBWeGfPJgpKOyDFIoenZsGppWrsc3TaWhwMWtZ9EmnH3x+YH7uWTQTDG+pIRAZ8 rWsrIQkZBTGtOzS8Ro1l4tdj1zh6AQjFa/Ub6jREICnPiI+UwfjDCLALMBGvBoO+CIE5 e4f/R0SJBIsJkhiBg92MDk8jfkQqFOtLiaTZY= X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=H49uHC+ovBJNcWDLUzyV16ecJUPkdKoVHPVznFUY7Cs=; b=BCWk76S3oKG7e09cssvSlYvyB9Mlw1p9duWRT/gdFAqND/vO7rDenaGjm6j07Rc1oB dGa8l/T5Pm2Mc1TJuAadPRbLaztf8GWhEmP+EcN1i381WhbO5TgItq7WaonVsSyPctei SYn8t1NcPIVqyot15G8Pl5dvv4iBbIKiy0WeoqX+au4lsdYbdjECf69f/QGGVxEi3HIn vG6fvoQAolLFPyEG99DqcczmrB7xxyP57b3qNuCET6vYmDVqBFhHuQLWfAtbqVpU7xKi +iGWPaar0CwOzbhVOEexk+zaSkakpWLOjsZxT5f4kxkgnWfGMwUBcgiFBhbp3m1KgbQ1 cj7g== X-Gm-Message-State: AOAM533g7d0fLBAPkQHoTR3cFeHvhJG16PJxPyAV5hxNSySWAy9Z48yw rH33Da7Ubuj1lqmeNdBWXFcGbIFpOypqses7znXmlGpzLflS X-Google-Smtp-Source: ABdhPJxlyJP6v2PpTTg6jps5cDhWRihQptlqkXeHaRccrvCqnyQWLtia9jrhT/lFApyaehS+EuCdQhuA9yvY X-Received: by 2002:a9d:1c90:: with SMTP id l16mr11472255ota.192.1602302853665; Fri, 09 Oct 2020 21:07:33 -0700 (PDT) Received: from localhost.localdomain ([192.19.223.252]) by smtp-relay.gmail.com with ESMTPS id 67sm1555892oop.24.2020.10.09.21.07.32 (version=TLS1_2 cipher=ECDHE-ECDSA-AES128-GCM-SHA256 bits=128/128); Fri, 09 Oct 2020 21:07:33 -0700 (PDT) X-Relaying-Domain: broadcom.com From: Ajit Khaparde To: dev@dpdk.org Cc: Venkat Duvvuru , stable@dpdk.org, Somnath Kotur Date: Fri, 9 Oct 2020 21:05:13 -0700 Message-Id: <20201010040520.63049-5-ajit.khaparde@broadcom.com> X-Mailer: git-send-email 2.21.1 (Apple Git-122.3) In-Reply-To: <20201009111130.10422-1-somnath.kotur@broadcom.com> References: <20201009111130.10422-1-somnath.kotur@broadcom.com> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Subject: [dpdk-stable] [PATCH v2 04/12] net/bnxt: fix PMD PF support in SR-IOV mode X-BeenThere: stable@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: patches for DPDK stable branches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: stable-bounces@dpdk.org Sender: "stable" From: Venkat Duvvuru 1. Implement HWRM_FUNC_VF_RESOURCE_CFG command and use it to reserve resources for VFs when NEW RM is enabled. 2. Invoke PF’s FUNC_CFG before configuring VFs resources. 3. Don’t consider max_rx_em_flows in max_l2_ctx calculation when VFs are configured. 4. Issue HWRM_FUNC_QCFG instead of HWRM_FUNC_QCAPS to find out the actual allocated resources for VF. 5. Don’t add random mac to the VF. 6. Handle completion type CMPL_BASE_TYPE_HWRM_FWD_REQ instead of CMPL_BASE_TYPE_HWRM_FWD_RESP. 7. Don't enable HWRM_FUNC_DRV_RGTR_INPUT_FLAGS_FWD_NONE_MODE when the list of HWRM commands that needs to be forwarded to the PF is specified in HWRM_FUNC_DRV_RGTR. 8. Update the HWRM commands list that can be forwared to the PF. Fixes: b7778e8a1c00 ("net/bnxt: refactor to properly allocate resources for PF/VF") Cc: stable@dpdk.org Signed-off-by: Venkat Duvvuru Reviewed-by: Somnath Kotur Reviewed-by: Ajit Khaparde --- drivers/net/bnxt/bnxt.h | 6 +- drivers/net/bnxt/bnxt_cpr.c | 6 +- drivers/net/bnxt/bnxt_ethdev.c | 40 +-- drivers/net/bnxt/bnxt_hwrm.c | 461 ++++++++++++++++++++------------- drivers/net/bnxt/bnxt_hwrm.h | 12 +- 5 files changed, 309 insertions(+), 216 deletions(-) diff --git a/drivers/net/bnxt/bnxt.h b/drivers/net/bnxt/bnxt.h index eca74486e..a951bca7a 100644 --- a/drivers/net/bnxt/bnxt.h +++ b/drivers/net/bnxt/bnxt.h @@ -167,6 +167,9 @@ #define BNXT_DEFAULT_VNIC_CHANGE_VF_ID_SFT \ HWRM_ASYNC_EVENT_CMPL_DEFAULT_VNIC_CHANGE_EVENT_DATA1_VF_ID_SFT +#define BNXT_HWRM_CMD_TO_FORWARD(cmd) \ + (bp->pf->vf_req_fwd[(cmd) / 32] |= (1 << ((cmd) % 32))) + struct bnxt_led_info { uint8_t num_leds; uint8_t led_id; @@ -664,9 +667,10 @@ struct bnxt { #define BNXT_FW_CAP_IF_CHANGE BIT(1) #define BNXT_FW_CAP_ERROR_RECOVERY BIT(2) #define BNXT_FW_CAP_ERR_RECOVER_RELOAD BIT(3) +#define BNXT_FW_CAP_HCOMM_FW_STATUS BIT(4) #define BNXT_FW_CAP_ADV_FLOW_MGMT BIT(5) #define BNXT_FW_CAP_ADV_FLOW_COUNTERS BIT(6) -#define BNXT_FW_CAP_HCOMM_FW_STATUS BIT(7) +#define BNXT_FW_CAP_LINK_ADMIN BIT(7) pthread_mutex_t flow_lock; diff --git a/drivers/net/bnxt/bnxt_cpr.c b/drivers/net/bnxt/bnxt_cpr.c index a3a7e6ab7..54923948f 100644 --- a/drivers/net/bnxt/bnxt_cpr.c +++ b/drivers/net/bnxt/bnxt_cpr.c @@ -239,7 +239,7 @@ void bnxt_handle_fwd_req(struct bnxt *bp, struct cmpl_base *cmpl) goto reject; } - if (bnxt_rcv_msg_from_vf(bp, vf_id, fwd_cmd) == true) { + if (bnxt_rcv_msg_from_vf(bp, vf_id, fwd_cmd)) { /* * In older firmware versions, the MAC had to be all zeros for * the VF to set it's MAC via hwrm_func_vf_cfg. Set to all @@ -254,6 +254,7 @@ void bnxt_handle_fwd_req(struct bnxt *bp, struct cmpl_base *cmpl) (const uint8_t *)"\x00\x00\x00\x00\x00"); } } + if (fwd_cmd->req_type == HWRM_CFA_L2_SET_RX_MASK) { struct hwrm_cfa_l2_set_rx_mask_input *srm = (void *)fwd_cmd; @@ -265,6 +266,7 @@ void bnxt_handle_fwd_req(struct bnxt *bp, struct cmpl_base *cmpl) HWRM_CFA_L2_SET_RX_MASK_INPUT_MASK_VLAN_NONVLAN | HWRM_CFA_L2_SET_RX_MASK_INPUT_MASK_ANYVLAN_NONVLAN); } + /* Forward */ rc = bnxt_hwrm_exec_fwd_resp(bp, fw_vf_id, fwd_cmd, req_len); if (rc) { @@ -306,7 +308,7 @@ int bnxt_event_hwrm_resp_handler(struct bnxt *bp, struct cmpl_base *cmp) bnxt_handle_async_event(bp, cmp); evt = 1; break; - case CMPL_BASE_TYPE_HWRM_FWD_RESP: + case CMPL_BASE_TYPE_HWRM_FWD_REQ: /* Handle HWRM forwarded responses */ bnxt_handle_fwd_req(bp, cmp); evt = 1; diff --git a/drivers/net/bnxt/bnxt_ethdev.c b/drivers/net/bnxt/bnxt_ethdev.c index 8b63134c3..b4654ec6a 100644 --- a/drivers/net/bnxt/bnxt_ethdev.c +++ b/drivers/net/bnxt/bnxt_ethdev.c @@ -5208,37 +5208,14 @@ static void bnxt_config_vf_req_fwd(struct bnxt *bp) if (!BNXT_PF(bp)) return; -#define ALLOW_FUNC(x) \ - { \ - uint32_t arg = (x); \ - bp->pf->vf_req_fwd[((arg) >> 5)] &= \ - ~rte_cpu_to_le_32(1 << ((arg) & 0x1f)); \ - } - - /* Forward all requests if firmware is new enough */ - if (((bp->fw_ver >= ((20 << 24) | (6 << 16) | (100 << 8))) && - (bp->fw_ver < ((20 << 24) | (7 << 16)))) || - ((bp->fw_ver >= ((20 << 24) | (8 << 16))))) { - memset(bp->pf->vf_req_fwd, 0xff, sizeof(bp->pf->vf_req_fwd)); - } else { - PMD_DRV_LOG(WARNING, - "Firmware too old for VF mailbox functionality\n"); - memset(bp->pf->vf_req_fwd, 0, sizeof(bp->pf->vf_req_fwd)); - } + memset(bp->pf->vf_req_fwd, 0, sizeof(bp->pf->vf_req_fwd)); - /* - * The following are used for driver cleanup. If we disallow these, - * VF drivers can't clean up cleanly. - */ - ALLOW_FUNC(HWRM_FUNC_DRV_UNRGTR); - ALLOW_FUNC(HWRM_VNIC_FREE); - ALLOW_FUNC(HWRM_RING_FREE); - ALLOW_FUNC(HWRM_RING_GRP_FREE); - ALLOW_FUNC(HWRM_VNIC_RSS_COS_LB_CTX_FREE); - ALLOW_FUNC(HWRM_CFA_L2_FILTER_FREE); - ALLOW_FUNC(HWRM_STAT_CTX_FREE); - ALLOW_FUNC(HWRM_PORT_PHY_QCFG); - ALLOW_FUNC(HWRM_VNIC_TPA_CFG); + if (!(bp->fw_cap & BNXT_FW_CAP_LINK_ADMIN)) + BNXT_HWRM_CMD_TO_FORWARD(HWRM_PORT_PHY_QCFG); + BNXT_HWRM_CMD_TO_FORWARD(HWRM_FUNC_CFG); + BNXT_HWRM_CMD_TO_FORWARD(HWRM_FUNC_VF_CFG); + BNXT_HWRM_CMD_TO_FORWARD(HWRM_CFA_L2_FILTER_ALLOC); + BNXT_HWRM_CMD_TO_FORWARD(HWRM_OEM_CMD); } uint16_t @@ -6189,7 +6166,10 @@ bnxt_uninit_resources(struct bnxt *bp, bool reconfig_dev) bnxt_free_int(bp); bnxt_free_mem(bp, reconfig_dev); + bnxt_hwrm_func_buf_unrgtr(bp); + rte_free(bp->pf->vf_req_buf); + rc = bnxt_hwrm_func_driver_unregister(bp, 0); bp->flags &= ~BNXT_FLAG_REGISTERED; bnxt_free_ctx_mem(bp); diff --git a/drivers/net/bnxt/bnxt_hwrm.c b/drivers/net/bnxt/bnxt_hwrm.c index faeaf4b5d..8133afc74 100644 --- a/drivers/net/bnxt/bnxt_hwrm.c +++ b/drivers/net/bnxt/bnxt_hwrm.c @@ -765,7 +765,7 @@ static int __bnxt_hwrm_func_qcaps(struct bnxt *bp) bp->first_vf_id = rte_le_to_cpu_16(resp->first_vf_id); bp->max_rx_em_flows = rte_le_to_cpu_16(resp->max_rx_em_flows); bp->max_l2_ctx = rte_le_to_cpu_16(resp->max_l2_ctxs); - if (!BNXT_CHIP_THOR(bp)) + if (!BNXT_CHIP_THOR(bp) && !bp->pdev->max_vfs) bp->max_l2_ctx += bp->max_rx_em_flows; /* TODO: For now, do not support VMDq/RFS on VFs. */ if (BNXT_PF(bp)) { @@ -803,6 +803,9 @@ static int __bnxt_hwrm_func_qcaps(struct bnxt *bp) if (flags & HWRM_FUNC_QCAPS_OUTPUT_FLAGS_HOT_RESET_CAPABLE) bp->fw_cap |= BNXT_FW_CAP_HOT_RESET; + if (flags & HWRM_FUNC_QCAPS_OUTPUT_FLAGS_LINK_ADMIN_STATUS_SUPPORTED) + bp->fw_cap |= BNXT_FW_CAP_LINK_ADMIN; + HWRM_UNLOCK(); return rc; @@ -818,16 +821,15 @@ int bnxt_hwrm_func_qcaps(struct bnxt *bp) if (rc) return rc; + /* On older FW, + * bnxt_hwrm_func_resc_qcaps can fail and cause init failure. + * But the error can be ignored. Return success. + */ rc = bnxt_hwrm_func_resc_qcaps(bp); if (!rc) bp->flags |= BNXT_FLAG_NEW_RM; } - /* On older FW, - * bnxt_hwrm_func_resc_qcaps can fail and cause init failure. - * But the error can be ignored. Return success. - */ - return 0; } @@ -916,14 +918,6 @@ int bnxt_hwrm_func_driver_register(struct bnxt *bp) memcpy(req.vf_req_fwd, bp->pf->vf_req_fwd, RTE_MIN(sizeof(req.vf_req_fwd), sizeof(bp->pf->vf_req_fwd))); - - /* - * PF can sniff HWRM API issued by VF. This can be set up by - * linux driver and inherited by the DPDK PF driver. Clear - * this HWRM sniffer list in FW because DPDK PF driver does - * not support this. - */ - flags |= HWRM_FUNC_DRV_RGTR_INPUT_FLAGS_FWD_NONE_MODE; } req.flags = rte_cpu_to_le_32(flags); @@ -1052,21 +1046,19 @@ int bnxt_hwrm_func_resc_qcaps(struct bnxt *bp) HWRM_CHECK_RESULT_SILENT(); - if (BNXT_VF(bp)) { - bp->max_rsscos_ctx = rte_le_to_cpu_16(resp->max_rsscos_ctx); - bp->max_cp_rings = rte_le_to_cpu_16(resp->max_cmpl_rings); - bp->max_tx_rings = rte_le_to_cpu_16(resp->max_tx_rings); - bp->max_rx_rings = rte_le_to_cpu_16(resp->max_rx_rings); - bp->max_ring_grps = rte_le_to_cpu_32(resp->max_hw_ring_grps); - /* func_resource_qcaps does not return max_rx_em_flows. - * So use the value provided by func_qcaps. - */ - bp->max_l2_ctx = rte_le_to_cpu_16(resp->max_l2_ctxs); - if (!BNXT_CHIP_THOR(bp)) - bp->max_l2_ctx += bp->max_rx_em_flows; - bp->max_vnics = rte_le_to_cpu_16(resp->max_vnics); - bp->max_stat_ctx = rte_le_to_cpu_16(resp->max_stat_ctx); - } + bp->max_rsscos_ctx = rte_le_to_cpu_16(resp->max_rsscos_ctx); + bp->max_cp_rings = rte_le_to_cpu_16(resp->max_cmpl_rings); + bp->max_tx_rings = rte_le_to_cpu_16(resp->max_tx_rings); + bp->max_rx_rings = rte_le_to_cpu_16(resp->max_rx_rings); + bp->max_ring_grps = rte_le_to_cpu_32(resp->max_hw_ring_grps); + /* func_resource_qcaps does not return max_rx_em_flows. + * So use the value provided by func_qcaps. + */ + bp->max_l2_ctx = rte_le_to_cpu_16(resp->max_l2_ctxs); + if (!BNXT_CHIP_THOR(bp) && !bp->pdev->max_vfs) + bp->max_l2_ctx += bp->max_rx_em_flows; + bp->max_vnics = rte_le_to_cpu_16(resp->max_vnics); + bp->max_stat_ctx = rte_le_to_cpu_16(resp->max_stat_ctx); bp->max_nq_rings = rte_le_to_cpu_16(resp->max_msix); bp->vf_resv_strategy = rte_le_to_cpu_16(resp->vf_reservation_strategy); if (bp->vf_resv_strategy > @@ -3300,33 +3292,8 @@ int bnxt_hwrm_port_mac_qcfg(struct bnxt *bp) return 0; } -static void copy_func_cfg_to_qcaps(struct hwrm_func_cfg_input *fcfg, - struct hwrm_func_qcaps_output *qcaps) -{ - qcaps->max_rsscos_ctx = fcfg->num_rsscos_ctxs; - memcpy(qcaps->mac_address, fcfg->dflt_mac_addr, - sizeof(qcaps->mac_address)); - qcaps->max_l2_ctxs = fcfg->num_l2_ctxs; - qcaps->max_rx_rings = fcfg->num_rx_rings; - qcaps->max_tx_rings = fcfg->num_tx_rings; - qcaps->max_cmpl_rings = fcfg->num_cmpl_rings; - qcaps->max_stat_ctx = fcfg->num_stat_ctxs; - qcaps->max_vfs = 0; - qcaps->first_vf_id = 0; - qcaps->max_vnics = fcfg->num_vnics; - qcaps->max_decap_records = 0; - qcaps->max_encap_records = 0; - qcaps->max_tx_wm_flows = 0; - qcaps->max_tx_em_flows = 0; - qcaps->max_rx_wm_flows = 0; - qcaps->max_rx_em_flows = 0; - qcaps->max_flow_id = 0; - qcaps->max_mcast_filters = fcfg->num_mcast_filters; - qcaps->max_sp_tx_rings = 0; - qcaps->max_hw_ring_grps = fcfg->num_hw_ring_grps; -} - -static int bnxt_hwrm_pf_func_cfg(struct bnxt *bp, int tx_rings) +static int bnxt_hwrm_pf_func_cfg(struct bnxt *bp, + struct bnxt_pf_resource_info *pf_resc) { struct hwrm_func_cfg_input req = {0}; struct hwrm_func_cfg_output *resp = bp->hwrm_cmd_resp_addr; @@ -3345,7 +3312,8 @@ static int bnxt_hwrm_pf_func_cfg(struct bnxt *bp, int tx_rings) if (BNXT_HAS_RING_GRPS(bp)) { enables |= HWRM_FUNC_CFG_INPUT_ENABLES_NUM_HW_RING_GRPS; - req.num_hw_ring_grps = rte_cpu_to_le_16(bp->max_ring_grps); + req.num_hw_ring_grps = + rte_cpu_to_le_16(pf_resc->num_hw_ring_grps); } else if (BNXT_HAS_NQ(bp)) { enables |= HWRM_FUNC_CFG_INPUT_ENABLES_NUM_MSIX; req.num_msix = rte_cpu_to_le_16(bp->max_nq_rings); @@ -3354,12 +3322,12 @@ static int bnxt_hwrm_pf_func_cfg(struct bnxt *bp, int tx_rings) req.flags = rte_cpu_to_le_32(bp->pf->func_cfg_flags); req.mtu = rte_cpu_to_le_16(BNXT_MAX_MTU); req.mru = rte_cpu_to_le_16(BNXT_VNIC_MRU(bp->eth_dev->data->mtu)); - req.num_rsscos_ctxs = rte_cpu_to_le_16(bp->max_rsscos_ctx); - req.num_stat_ctxs = rte_cpu_to_le_16(bp->max_stat_ctx); - req.num_cmpl_rings = rte_cpu_to_le_16(bp->max_cp_rings); - req.num_tx_rings = rte_cpu_to_le_16(tx_rings); - req.num_rx_rings = rte_cpu_to_le_16(bp->max_rx_rings); - req.num_l2_ctxs = rte_cpu_to_le_16(bp->max_l2_ctx); + req.num_rsscos_ctxs = rte_cpu_to_le_16(pf_resc->num_rsscos_ctxs); + req.num_stat_ctxs = rte_cpu_to_le_16(pf_resc->num_stat_ctxs); + req.num_cmpl_rings = rte_cpu_to_le_16(pf_resc->num_cp_rings); + req.num_tx_rings = rte_cpu_to_le_16(pf_resc->num_tx_rings); + req.num_rx_rings = rte_cpu_to_le_16(pf_resc->num_rx_rings); + req.num_l2_ctxs = rte_cpu_to_le_16(pf_resc->num_l2_ctxs); req.num_vnics = rte_cpu_to_le_16(bp->max_vnics); req.fid = rte_cpu_to_le_16(0xffff); req.enables = rte_cpu_to_le_32(enables); @@ -3374,9 +3342,43 @@ static int bnxt_hwrm_pf_func_cfg(struct bnxt *bp, int tx_rings) return rc; } -static void populate_vf_func_cfg_req(struct bnxt *bp, - struct hwrm_func_cfg_input *req, - int num_vfs) +/* min values are the guaranteed resources and max values are subject + * to availability. The strategy for now is to keep both min & max + * values the same. + */ +static void +bnxt_fill_vf_func_cfg_req_new(struct bnxt *bp, + struct hwrm_func_vf_resource_cfg_input *req, + int num_vfs) +{ + req->max_rsscos_ctx = rte_cpu_to_le_16(bp->max_rsscos_ctx / + (num_vfs + 1)); + req->min_rsscos_ctx = req->max_rsscos_ctx; + req->max_stat_ctx = rte_cpu_to_le_16(bp->max_stat_ctx / (num_vfs + 1)); + req->min_stat_ctx = req->max_stat_ctx; + req->max_cmpl_rings = rte_cpu_to_le_16(bp->max_cp_rings / + (num_vfs + 1)); + req->min_cmpl_rings = req->max_cmpl_rings; + req->max_tx_rings = rte_cpu_to_le_16(bp->max_tx_rings / (num_vfs + 1)); + req->min_tx_rings = req->max_tx_rings; + req->max_rx_rings = rte_cpu_to_le_16(bp->max_rx_rings / (num_vfs + 1)); + req->min_rx_rings = req->max_rx_rings; + req->max_l2_ctxs = rte_cpu_to_le_16(bp->max_l2_ctx / (num_vfs + 1)); + req->min_l2_ctxs = req->max_l2_ctxs; + /* TODO: For now, do not support VMDq/RFS on VFs. */ + req->max_vnics = rte_cpu_to_le_16(1); + req->min_vnics = req->max_vnics; + req->max_hw_ring_grps = rte_cpu_to_le_16(bp->max_ring_grps / + (num_vfs + 1)); + req->min_hw_ring_grps = req->max_hw_ring_grps; + req->flags = + rte_cpu_to_le_16(HWRM_FUNC_VF_RESOURCE_CFG_INPUT_FLAGS_MIN_GUARANTEED); +} + +static void +bnxt_fill_vf_func_cfg_req_old(struct bnxt *bp, + struct hwrm_func_cfg_input *req, + int num_vfs) { req->enables = rte_cpu_to_le_32(HWRM_FUNC_CFG_INPUT_ENABLES_MTU | HWRM_FUNC_CFG_INPUT_ENABLES_MRU | @@ -3407,60 +3409,29 @@ static void populate_vf_func_cfg_req(struct bnxt *bp, (num_vfs + 1)); } -static void add_random_mac_if_needed(struct bnxt *bp, - struct hwrm_func_cfg_input *cfg_req, - int vf) -{ - struct rte_ether_addr mac; - - if (bnxt_hwrm_func_qcfg_vf_default_mac(bp, vf, &mac)) - return; - - if (memcmp(mac.addr_bytes, "\x00\x00\x00\x00\x00", 6) == 0) { - cfg_req->enables |= - rte_cpu_to_le_32(HWRM_FUNC_CFG_INPUT_ENABLES_DFLT_MAC_ADDR); - rte_eth_random_addr(cfg_req->dflt_mac_addr); - bp->pf->vf_info[vf].random_mac = true; - } else { - memcpy(cfg_req->dflt_mac_addr, mac.addr_bytes, - RTE_ETHER_ADDR_LEN); - } -} - -static int reserve_resources_from_vf(struct bnxt *bp, - struct hwrm_func_cfg_input *cfg_req, +/* Update the port wide resource values based on how many resources + * got allocated to the VF. + */ +static int bnxt_update_max_resources(struct bnxt *bp, int vf) { - struct hwrm_func_qcaps_input req = {0}; - struct hwrm_func_qcaps_output *resp = bp->hwrm_cmd_resp_addr; + struct hwrm_func_qcfg_input req = {0}; + struct hwrm_func_qcfg_output *resp = bp->hwrm_cmd_resp_addr; int rc; /* Get the actual allocated values now */ - HWRM_PREP(&req, HWRM_FUNC_QCAPS, BNXT_USE_CHIMP_MB); + HWRM_PREP(&req, HWRM_FUNC_QCFG, BNXT_USE_CHIMP_MB); req.fid = rte_cpu_to_le_16(bp->pf->vf_info[vf].fid); rc = bnxt_hwrm_send_message(bp, &req, sizeof(req), BNXT_USE_CHIMP_MB); + HWRM_CHECK_RESULT(); - if (rc) { - PMD_DRV_LOG(ERR, "hwrm_func_qcaps failed rc:%d\n", rc); - copy_func_cfg_to_qcaps(cfg_req, resp); - } else if (resp->error_code) { - rc = rte_le_to_cpu_16(resp->error_code); - PMD_DRV_LOG(ERR, "hwrm_func_qcaps error %d\n", rc); - copy_func_cfg_to_qcaps(cfg_req, resp); - } - - bp->max_rsscos_ctx -= rte_le_to_cpu_16(resp->max_rsscos_ctx); - bp->max_stat_ctx -= rte_le_to_cpu_16(resp->max_stat_ctx); - bp->max_cp_rings -= rte_le_to_cpu_16(resp->max_cmpl_rings); - bp->max_tx_rings -= rte_le_to_cpu_16(resp->max_tx_rings); - bp->max_rx_rings -= rte_le_to_cpu_16(resp->max_rx_rings); - bp->max_l2_ctx -= rte_le_to_cpu_16(resp->max_l2_ctxs); - /* - * TODO: While not supporting VMDq with VFs, max_vnics is always - * forced to 1 in this case - */ - //bp->max_vnics -= rte_le_to_cpu_16(esp->max_vnics); - bp->max_ring_grps -= rte_le_to_cpu_16(resp->max_hw_ring_grps); + bp->max_rsscos_ctx -= rte_le_to_cpu_16(resp->alloc_rsscos_ctx); + bp->max_stat_ctx -= rte_le_to_cpu_16(resp->alloc_stat_ctx); + bp->max_cp_rings -= rte_le_to_cpu_16(resp->alloc_cmpl_rings); + bp->max_tx_rings -= rte_le_to_cpu_16(resp->alloc_tx_rings); + bp->max_rx_rings -= rte_le_to_cpu_16(resp->alloc_rx_rings); + bp->max_l2_ctx -= rte_le_to_cpu_16(resp->alloc_l2_ctx); + bp->max_ring_grps -= rte_le_to_cpu_16(resp->alloc_hw_ring_grps); HWRM_UNLOCK(); @@ -3485,7 +3456,8 @@ int bnxt_hwrm_func_qcfg_current_vf_vlan(struct bnxt *bp, int vf) return rc; } -static int update_pf_resource_max(struct bnxt *bp) +static int bnxt_query_pf_resources(struct bnxt *bp, + struct bnxt_pf_resource_info *pf_resc) { struct hwrm_func_qcfg_input req = {0}; struct hwrm_func_qcfg_output *resp = bp->hwrm_cmd_resp_addr; @@ -3497,8 +3469,13 @@ static int update_pf_resource_max(struct bnxt *bp) rc = bnxt_hwrm_send_message(bp, &req, sizeof(req), BNXT_USE_CHIMP_MB); HWRM_CHECK_RESULT(); - /* Only TX ring value reflects actual allocation? TODO */ - bp->max_tx_rings = rte_le_to_cpu_16(resp->alloc_tx_rings); + pf_resc->num_tx_rings = rte_le_to_cpu_16(resp->alloc_tx_rings); + pf_resc->num_rsscos_ctxs = rte_le_to_cpu_16(resp->alloc_rsscos_ctx); + pf_resc->num_stat_ctxs = rte_le_to_cpu_16(resp->alloc_stat_ctx); + pf_resc->num_cp_rings = rte_le_to_cpu_16(resp->alloc_cmpl_rings); + pf_resc->num_rx_rings = rte_le_to_cpu_16(resp->alloc_rx_rings); + pf_resc->num_l2_ctxs = rte_le_to_cpu_16(resp->alloc_l2_ctx); + pf_resc->num_hw_ring_grps = rte_le_to_cpu_32(resp->alloc_hw_ring_grps); bp->pf->evb_mode = resp->evb_mode; HWRM_UNLOCK(); @@ -3506,8 +3483,42 @@ static int update_pf_resource_max(struct bnxt *bp) return rc; } +static void +bnxt_calculate_pf_resources(struct bnxt *bp, + struct bnxt_pf_resource_info *pf_resc, + int num_vfs) +{ + if (!num_vfs) { + pf_resc->num_rsscos_ctxs = bp->max_rsscos_ctx; + pf_resc->num_stat_ctxs = bp->max_stat_ctx; + pf_resc->num_cp_rings = bp->max_cp_rings; + pf_resc->num_tx_rings = bp->max_tx_rings; + pf_resc->num_rx_rings = bp->max_rx_rings; + pf_resc->num_l2_ctxs = bp->max_l2_ctx; + pf_resc->num_hw_ring_grps = bp->max_ring_grps; + + return; + } + + pf_resc->num_rsscos_ctxs = bp->max_rsscos_ctx / (num_vfs + 1) + + bp->max_rsscos_ctx % (num_vfs + 1); + pf_resc->num_stat_ctxs = bp->max_stat_ctx / (num_vfs + 1) + + bp->max_stat_ctx % (num_vfs + 1); + pf_resc->num_cp_rings = bp->max_cp_rings / (num_vfs + 1) + + bp->max_cp_rings % (num_vfs + 1); + pf_resc->num_tx_rings = bp->max_tx_rings / (num_vfs + 1) + + bp->max_tx_rings % (num_vfs + 1); + pf_resc->num_rx_rings = bp->max_rx_rings / (num_vfs + 1) + + bp->max_rx_rings % (num_vfs + 1); + pf_resc->num_l2_ctxs = bp->max_l2_ctx / (num_vfs + 1) + + bp->max_l2_ctx % (num_vfs + 1); + pf_resc->num_hw_ring_grps = bp->max_ring_grps / (num_vfs + 1) + + bp->max_ring_grps % (num_vfs + 1); +} + int bnxt_hwrm_allocate_pf_only(struct bnxt *bp) { + struct bnxt_pf_resource_info pf_resc = { 0 }; int rc; if (!BNXT_PF(bp)) { @@ -3519,82 +3530,100 @@ int bnxt_hwrm_allocate_pf_only(struct bnxt *bp) if (rc) return rc; + bnxt_calculate_pf_resources(bp, &pf_resc, 0); + bp->pf->func_cfg_flags &= ~(HWRM_FUNC_CFG_INPUT_FLAGS_STD_TX_RING_MODE_ENABLE | HWRM_FUNC_CFG_INPUT_FLAGS_STD_TX_RING_MODE_DISABLE); bp->pf->func_cfg_flags |= HWRM_FUNC_CFG_INPUT_FLAGS_STD_TX_RING_MODE_DISABLE; - rc = bnxt_hwrm_pf_func_cfg(bp, bp->max_tx_rings); + rc = bnxt_hwrm_pf_func_cfg(bp, &pf_resc); rc = __bnxt_hwrm_func_qcaps(bp); return rc; } -int bnxt_hwrm_allocate_vfs(struct bnxt *bp, int num_vfs) +static int +bnxt_configure_vf_req_buf(struct bnxt *bp, int num_vfs) { - struct hwrm_func_cfg_input req = {0}; - struct hwrm_func_cfg_output *resp = bp->hwrm_cmd_resp_addr; - int i; - size_t sz; - int rc = 0; - size_t req_buf_sz; - - if (!BNXT_PF(bp)) { - PMD_DRV_LOG(ERR, "Attempt to allcoate VFs on a VF!\n"); - return -EINVAL; - } - - rc = bnxt_hwrm_func_qcaps(bp); - - if (rc) - return rc; - - bp->pf->active_vfs = num_vfs; - - /* - * First, configure the PF to only use one TX ring. This ensures that - * there are enough rings for all VFs. - * - * If we don't do this, when we call func_alloc() later, we will lock - * extra rings to the PF that won't be available during func_cfg() of - * the VFs. - * - * This has been fixed with firmware versions above 20.6.54 - */ - bp->pf->func_cfg_flags &= - ~(HWRM_FUNC_CFG_INPUT_FLAGS_STD_TX_RING_MODE_ENABLE | - HWRM_FUNC_CFG_INPUT_FLAGS_STD_TX_RING_MODE_DISABLE); - bp->pf->func_cfg_flags |= - HWRM_FUNC_CFG_INPUT_FLAGS_STD_TX_RING_MODE_ENABLE; - rc = bnxt_hwrm_pf_func_cfg(bp, 1); - if (rc) - return rc; + size_t req_buf_sz, sz; + int i, rc; - /* - * Now, create and register a buffer to hold forwarded VF requests - */ req_buf_sz = num_vfs * HWRM_MAX_REQ_LEN; bp->pf->vf_req_buf = rte_malloc("bnxt_vf_fwd", req_buf_sz, page_roundup(num_vfs * HWRM_MAX_REQ_LEN)); if (bp->pf->vf_req_buf == NULL) { - rc = -ENOMEM; - goto error_free; + return -ENOMEM; } + for (sz = 0; sz < req_buf_sz; sz += getpagesize()) rte_mem_lock_page(((char *)bp->pf->vf_req_buf) + sz); + for (i = 0; i < num_vfs; i++) bp->pf->vf_info[i].req_buf = ((char *)bp->pf->vf_req_buf) + - (i * HWRM_MAX_REQ_LEN); + (i * HWRM_MAX_REQ_LEN); - rc = bnxt_hwrm_func_buf_rgtr(bp); + rc = bnxt_hwrm_func_buf_rgtr(bp, num_vfs); if (rc) - goto error_free; + rte_free(bp->pf->vf_req_buf); + + return rc; +} - populate_vf_func_cfg_req(bp, &req, num_vfs); +static int +bnxt_process_vf_resc_config_new(struct bnxt *bp, int num_vfs) +{ + struct hwrm_func_vf_resource_cfg_output *resp = bp->hwrm_cmd_resp_addr; + struct hwrm_func_vf_resource_cfg_input req = {0}; + int i, rc = 0; + bnxt_fill_vf_func_cfg_req_new(bp, &req, num_vfs); bp->pf->active_vfs = 0; for (i = 0; i < num_vfs; i++) { - add_random_mac_if_needed(bp, &req, i); + HWRM_PREP(&req, HWRM_FUNC_VF_RESOURCE_CFG, BNXT_USE_CHIMP_MB); + req.vf_id = rte_cpu_to_le_16(bp->pf->vf_info[i].fid); + rc = bnxt_hwrm_send_message(bp, + &req, + sizeof(req), + BNXT_USE_CHIMP_MB); + if (rc || resp->error_code) { + PMD_DRV_LOG(ERR, + "Failed to initialize VF %d\n", i); + PMD_DRV_LOG(ERR, + "Not all VFs available. (%d, %d)\n", + rc, resp->error_code); + HWRM_UNLOCK(); + + /* If the first VF configuration itself fails, + * unregister the vf_fwd_request buffer. + */ + if (i == 0) + bnxt_hwrm_func_buf_unrgtr(bp); + break; + } + HWRM_UNLOCK(); + + /* Update the max resource values based on the resource values + * allocated to the VF. + */ + bnxt_update_max_resources(bp, i); + bp->pf->active_vfs++; + bnxt_hwrm_func_clr_stats(bp, bp->pf->vf_info[i].fid); + } + + return 0; +} + +static int +bnxt_process_vf_resc_config_old(struct bnxt *bp, int num_vfs) +{ + struct hwrm_func_cfg_output *resp = bp->hwrm_cmd_resp_addr; + struct hwrm_func_cfg_input req = {0}; + int i, rc; + bnxt_fill_vf_func_cfg_req_old(bp, &req, num_vfs); + + bp->pf->active_vfs = 0; + for (i = 0; i < num_vfs; i++) { HWRM_PREP(&req, HWRM_FUNC_CFG, BNXT_USE_CHIMP_MB); req.flags = rte_cpu_to_le_32(bp->pf->vf_info[i].func_cfg_flags); req.fid = rte_cpu_to_le_16(bp->pf->vf_info[i].fid); @@ -3609,40 +3638,107 @@ int bnxt_hwrm_allocate_vfs(struct bnxt *bp, int num_vfs) if (rc || resp->error_code) { PMD_DRV_LOG(ERR, - "Failed to initizlie VF %d\n", i); + "Failed to initialize VF %d\n", i); PMD_DRV_LOG(ERR, "Not all VFs available. (%d, %d)\n", rc, resp->error_code); HWRM_UNLOCK(); + + /* If the first VF configuration itself fails, + * unregister the vf_fwd_request buffer. + */ + if (i == 0) + bnxt_hwrm_func_buf_unrgtr(bp); break; } HWRM_UNLOCK(); - reserve_resources_from_vf(bp, &req, i); + /* Update the max resource values based on the resource values + * allocated to the VF. + */ + bnxt_update_max_resources(bp, i); bp->pf->active_vfs++; bnxt_hwrm_func_clr_stats(bp, bp->pf->vf_info[i].fid); } + return 0; +} + +static void +bnxt_configure_vf_resources(struct bnxt *bp, int num_vfs) +{ + if (bp->flags & BNXT_FLAG_NEW_RM) + bnxt_process_vf_resc_config_new(bp, num_vfs); + else + bnxt_process_vf_resc_config_old(bp, num_vfs); +} + +static void +bnxt_update_pf_resources(struct bnxt *bp, + struct bnxt_pf_resource_info *pf_resc) +{ + bp->max_rsscos_ctx = pf_resc->num_rsscos_ctxs; + bp->max_stat_ctx = pf_resc->num_stat_ctxs; + bp->max_cp_rings = pf_resc->num_cp_rings; + bp->max_tx_rings = pf_resc->num_tx_rings; + bp->max_rx_rings = pf_resc->num_rx_rings; + bp->max_ring_grps = pf_resc->num_hw_ring_grps; +} + +static int32_t +bnxt_configure_pf_resources(struct bnxt *bp, + struct bnxt_pf_resource_info *pf_resc) +{ /* - * Now configure the PF to use "the rest" of the resources - * We're using STD_TX_RING_MODE here though which will limit the TX - * rings. This will allow QoS to function properly. Not setting this + * We're using STD_TX_RING_MODE here which will limit the TX + * rings. This will allow QoS to function properly. Not setting this * will cause PF rings to break bandwidth settings. */ - rc = bnxt_hwrm_pf_func_cfg(bp, bp->max_tx_rings); + bp->pf->func_cfg_flags &= + ~(HWRM_FUNC_CFG_INPUT_FLAGS_STD_TX_RING_MODE_ENABLE | + HWRM_FUNC_CFG_INPUT_FLAGS_STD_TX_RING_MODE_DISABLE); + bp->pf->func_cfg_flags |= + HWRM_FUNC_CFG_INPUT_FLAGS_STD_TX_RING_MODE_ENABLE; + return bnxt_hwrm_pf_func_cfg(bp, pf_resc); +} + +int bnxt_hwrm_allocate_vfs(struct bnxt *bp, int num_vfs) +{ + struct bnxt_pf_resource_info pf_resc = { 0 }; + int rc; + + if (!BNXT_PF(bp)) { + PMD_DRV_LOG(ERR, "Attempt to allocate VFs on a VF!\n"); + return -EINVAL; + } + + rc = bnxt_hwrm_func_qcaps(bp); if (rc) - goto error_free; + return rc; + + bnxt_calculate_pf_resources(bp, &pf_resc, num_vfs); - rc = update_pf_resource_max(bp); + rc = bnxt_configure_pf_resources(bp, &pf_resc); if (rc) - goto error_free; + return rc; - return rc; + rc = bnxt_query_pf_resources(bp, &pf_resc); + if (rc) + return rc; -error_free: - bnxt_hwrm_func_buf_unrgtr(bp); - return rc; + /* + * Now, create and register a buffer to hold forwarded VF requests + */ + rc = bnxt_configure_vf_req_buf(bp, num_vfs); + if (rc) + return rc; + + bnxt_configure_vf_resources(bp, num_vfs); + + bnxt_update_pf_resources(bp, &pf_resc); + + return 0; } int bnxt_hwrm_pf_evb_mode(struct bnxt *bp) @@ -3747,23 +3843,24 @@ int bnxt_set_rx_mask_no_vlan(struct bnxt *bp, struct bnxt_vnic_info *vnic) return bnxt_hwrm_cfa_l2_set_rx_mask(bp, vnic, 0, NULL); } -int bnxt_hwrm_func_buf_rgtr(struct bnxt *bp) +int bnxt_hwrm_func_buf_rgtr(struct bnxt *bp, int num_vfs) { - int rc = 0; - struct hwrm_func_buf_rgtr_input req = {.req_type = 0 }; struct hwrm_func_buf_rgtr_output *resp = bp->hwrm_cmd_resp_addr; + struct hwrm_func_buf_rgtr_input req = {.req_type = 0 }; + int rc; HWRM_PREP(&req, HWRM_FUNC_BUF_RGTR, BNXT_USE_CHIMP_MB); req.req_buf_num_pages = rte_cpu_to_le_16(1); - req.req_buf_page_size = rte_cpu_to_le_16( - page_getenum(bp->pf->active_vfs * HWRM_MAX_REQ_LEN)); + req.req_buf_page_size = + rte_cpu_to_le_16(page_getenum(num_vfs * HWRM_MAX_REQ_LEN)); req.req_buf_len = rte_cpu_to_le_16(HWRM_MAX_REQ_LEN); req.req_buf_page_addr0 = rte_cpu_to_le_64(rte_malloc_virt2iova(bp->pf->vf_req_buf)); if (req.req_buf_page_addr0 == RTE_BAD_IOVA) { PMD_DRV_LOG(ERR, "unable to map buffer address to physical memory\n"); + HWRM_UNLOCK(); return -ENOMEM; } diff --git a/drivers/net/bnxt/bnxt_hwrm.h b/drivers/net/bnxt/bnxt_hwrm.h index e98b1fe41..a7fa7f66b 100644 --- a/drivers/net/bnxt/bnxt_hwrm.h +++ b/drivers/net/bnxt/bnxt_hwrm.h @@ -107,6 +107,16 @@ enum bnxt_flow_dir { BNXT_DIR_MAX }; +struct bnxt_pf_resource_info { + uint16_t num_rsscos_ctxs; + uint16_t num_stat_ctxs; + uint16_t num_tx_rings; + uint16_t num_rx_rings; + uint16_t num_cp_rings; + uint16_t num_l2_ctxs; + uint32_t num_hw_ring_grps; +}; + #define BNXT_CTX_VAL_INVAL 0xFFFF int bnxt_hwrm_cfa_l2_clear_rx_mask(struct bnxt *bp, @@ -127,7 +137,7 @@ int bnxt_hwrm_exec_fwd_resp(struct bnxt *bp, uint16_t target_id, int bnxt_hwrm_reject_fwd_resp(struct bnxt *bp, uint16_t target_id, void *encaped, size_t ec_size); -int bnxt_hwrm_func_buf_rgtr(struct bnxt *bp); +int bnxt_hwrm_func_buf_rgtr(struct bnxt *bp, int num_vfs); int bnxt_hwrm_func_buf_unrgtr(struct bnxt *bp); int bnxt_hwrm_func_driver_register(struct bnxt *bp); int bnxt_hwrm_func_qcaps(struct bnxt *bp); -- 2.21.1 (Apple Git-122.3)