From patchwork Tue Dec 20 09:13:25 2016 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Selvin Xavier X-Patchwork-Id: 9481289 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id F38EA600CA for ; Tue, 20 Dec 2016 09:15:08 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id DBFF9283E5 for ; Tue, 20 Dec 2016 09:15:08 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id D03C628476; Tue, 20 Dec 2016 09:15:08 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-6.8 required=2.0 tests=BAYES_00,DKIM_SIGNED, RCVD_IN_DNSWL_HI,T_DKIM_INVALID autolearn=ham version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id F3F9F283E5 for ; Tue, 20 Dec 2016 09:15:07 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1762333AbcLTJO6 (ORCPT ); Tue, 20 Dec 2016 04:14:58 -0500 Received: from mail-qt0-f176.google.com ([209.85.216.176]:36591 "EHLO mail-qt0-f176.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753307AbcLTJOx (ORCPT ); Tue, 20 Dec 2016 04:14:53 -0500 Received: by mail-qt0-f176.google.com with SMTP id w33so170963117qtc.3 for ; Tue, 20 Dec 2016 01:14:51 -0800 (PST) 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; bh=aQ+1v1Oo+23pwrzM0YtdgqrNsUskBW1BRQjn1OTm5aI=; b=Ex6GosRasoSNzHdZw8y2G0qUOPVnuaEnZIfn+iB5U9AfWG/6xfLWNJlTQPxbf3wlez fo7+I6k7f03SbNOvBTOga+ieOvlkfIhWlQCCeiotZWv48hOzWM7kdd4Kid5YtzcV9pzm ECjtgyRIzOLWij8oDHqDWqWr2dCGY0sJ1e1Zc= 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; bh=aQ+1v1Oo+23pwrzM0YtdgqrNsUskBW1BRQjn1OTm5aI=; b=T/JhuaxnY2v8aJrhFGLLdMkqX9nWV1yRz8fJ4IyABEn0vqODUxKB54ikj2H+rG+ZVg dIenY9yuhBciEWCVlv5HaTxncgWKbc3OJ1QaHegDpPxYtk0WnsvSGXNSchOkbffrCQYX /0C4SMpOJKoadKSdYGOJzQVH7wdL+hI3me531KvyQevhMpyfBceshjeVnB2jcoA+Ie/e LWrSn9uHQsnjBz3KDoVof8p6JwY68WSsKf7iwGvkrUnODm4s87wD7GyijHuFD4Zcg3ua M6Gvx8532MVk1LxfSj5jEM7/RuPYbaoMHfTdwz8y9jzo076IrhXPLpCNRr8eE6MfCKHu wpfw== X-Gm-Message-State: AIkVDXL1yw3YXIh/sYjyxl6OAy+x/fmeLXJAdDPaD4KBZIZNVPaL/8jij3SeG1743tVg110v X-Received: by 10.200.43.82 with SMTP id 18mr18674738qtv.63.1482225290564; Tue, 20 Dec 2016 01:14:50 -0800 (PST) Received: from dhcp-10-192-206-197.iig.avagotech.net ([192.19.239.250]) by smtp.gmail.com with ESMTPSA id b63sm12494452qka.39.2016.12.20.01.14.46 (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Tue, 20 Dec 2016 01:14:50 -0800 (PST) From: Selvin Xavier To: dledford@redhat.com, linux-rdma@vger.kernel.org Cc: netdev@vger.kernel.org, michael.chan@broadcom.com, Selvin Xavier , Eddie Wai , Devesh Sharma , Somnath Kotur , Sriharsha Basavapatna Subject: [PATCH for bnxt_re V3 15/21] bnxt_re: Support post_recv Date: Tue, 20 Dec 2016 01:13:25 -0800 Message-Id: <1482225211-22423-16-git-send-email-selvin.xavier@broadcom.com> X-Mailer: git-send-email 2.5.5 In-Reply-To: <1482225211-22423-1-git-send-email-selvin.xavier@broadcom.com> References: <1482225211-22423-1-git-send-email-selvin.xavier@broadcom.com> Sender: linux-rdma-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-rdma@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP Enables the fastpath verb ib_post_recv. v3: Fixes sparse warnings Signed-off-by: Eddie Wai Signed-off-by: Devesh Sharma Signed-off-by: Somnath Kotur Signed-off-by: Sriharsha Basavapatna Signed-off-by: Selvin Xavier --- drivers/infiniband/hw/bnxtre/bnxt_qplib_fp.c | 100 +++++++++++++++++++ drivers/infiniband/hw/bnxtre/bnxt_qplib_fp.h | 8 ++ drivers/infiniband/hw/bnxtre/bnxt_re_ib_verbs.c | 123 ++++++++++++++++++++++++ drivers/infiniband/hw/bnxtre/bnxt_re_ib_verbs.h | 2 + drivers/infiniband/hw/bnxtre/bnxt_re_main.c | 2 + 5 files changed, 235 insertions(+) diff --git a/drivers/infiniband/hw/bnxtre/bnxt_qplib_fp.c b/drivers/infiniband/hw/bnxtre/bnxt_qplib_fp.c index d1ccf66..eeafb2a 100644 --- a/drivers/infiniband/hw/bnxtre/bnxt_qplib_fp.c +++ b/drivers/infiniband/hw/bnxtre/bnxt_qplib_fp.c @@ -1097,6 +1097,37 @@ void *bnxt_qplib_get_qp1_sq_buf(struct bnxt_qplib_qp *qp, return NULL; } +u32 bnxt_qplib_get_rq_prod_index(struct bnxt_qplib_qp *qp) +{ + struct bnxt_qplib_q *rq = &qp->rq; + + return HWQ_CMP(rq->hwq.prod, &rq->hwq); +} + +dma_addr_t bnxt_qplib_get_qp_buf_from_index(struct bnxt_qplib_qp *qp, u32 index) +{ + return (qp->rq_hdr_buf_map + index * qp->rq_hdr_buf_size); +} + +void *bnxt_qplib_get_qp1_rq_buf(struct bnxt_qplib_qp *qp, + struct bnxt_qplib_sge *sge) +{ + struct bnxt_qplib_q *rq = &qp->rq; + u32 sw_prod; + + memset(sge, 0, sizeof(*sge)); + + if (qp->rq_hdr_buf) { + sw_prod = HWQ_CMP(rq->hwq.prod, &rq->hwq); + sge->addr = (dma_addr_t)(qp->rq_hdr_buf_map + + sw_prod * qp->rq_hdr_buf_size); + sge->lkey = 0xFFFFFFFF; + sge->size = qp->rq_hdr_buf_size; + return qp->rq_hdr_buf + sw_prod * sge->size; + } + return NULL; +} + void bnxt_qplib_post_send_db(struct bnxt_qplib_qp *qp) { struct bnxt_qplib_q *sq = &qp->sq; @@ -1346,6 +1377,75 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp, return rc; } +void bnxt_qplib_post_recv_db(struct bnxt_qplib_qp *qp) +{ + struct bnxt_qplib_q *rq = &qp->rq; + struct dbr_dbr db_msg = { 0 }; + u32 sw_prod; + + sw_prod = HWQ_CMP(rq->hwq.prod, &rq->hwq); + db_msg.index = cpu_to_le32((sw_prod << DBR_DBR_INDEX_SFT) & + DBR_DBR_INDEX_MASK); + db_msg.type_xid = + cpu_to_le32(((qp->id << DBR_DBR_XID_SFT) & DBR_DBR_XID_MASK) | + DBR_DBR_TYPE_RQ); + + /* Flush the writes to HW Rx WQE before the ringing Rx DB */ + wmb(); + __iowrite64_copy(qp->dpi->dbr, &db_msg, sizeof(db_msg) / sizeof(u64)); +} + +int bnxt_qplib_post_recv(struct bnxt_qplib_qp *qp, + struct bnxt_qplib_swqe *wqe) +{ + struct bnxt_qplib_q *rq = &qp->rq; + struct rq_wqe *rqe, **rqe_ptr; + struct sq_sge *hw_sge; + u32 sw_prod; + int i, rc = 0; + + if (qp->state == CMDQ_MODIFY_QP_NEW_STATE_ERR) { + dev_err(&rq->hwq.pdev->dev, + "QPLIB: FP: QP (0x%x) is in the 0x%x state", + qp->id, qp->state); + rc = -EINVAL; + goto done; + } + if (HWQ_CMP((rq->hwq.prod + 1), &rq->hwq) == + HWQ_CMP(rq->hwq.cons, &rq->hwq)) { + dev_err(&rq->hwq.pdev->dev, + "QPLIB: FP: QP (0x%x) RQ is full!", qp->id); + rc = -EINVAL; + goto done; + } + sw_prod = HWQ_CMP(rq->hwq.prod, &rq->hwq); + rq->swq[sw_prod].wr_id = wqe->wr_id; + + rqe_ptr = (struct rq_wqe **)rq->hwq.pbl_ptr; + rqe = &rqe_ptr[RQE_PG(sw_prod)][RQE_IDX(sw_prod)]; + + memset(rqe, 0, BNXT_QPLIB_MAX_RQE_ENTRY_SIZE); + + /* Calculate wqe_size16 and data_len */ + for (i = 0, hw_sge = (struct sq_sge *)rqe->data; + i < wqe->num_sge; i++, hw_sge++) { + hw_sge->va_or_pa = cpu_to_le64(wqe->sg_list[i].addr); + hw_sge->l_key = cpu_to_le32(wqe->sg_list[i].lkey); + hw_sge->size = cpu_to_le32(wqe->sg_list[i].size); + } + rqe->wqe_type = wqe->type; + rqe->flags = wqe->flags; + rqe->wqe_size = wqe->num_sge + + ((offsetof(typeof(*rqe), data) + 15) >> 4); + + /* Supply the rqe->wr_id index to the wr_id_tbl for now */ + rqe->wr_id[0] = cpu_to_le32(sw_prod); + + rq->hwq.prod++; +done: + return rc; +} + /* CQ */ /* Spinlock must be held */ diff --git a/drivers/infiniband/hw/bnxtre/bnxt_qplib_fp.h b/drivers/infiniband/hw/bnxtre/bnxt_qplib_fp.h index 0a87920..e160050 100644 --- a/drivers/infiniband/hw/bnxtre/bnxt_qplib_fp.h +++ b/drivers/infiniband/hw/bnxtre/bnxt_qplib_fp.h @@ -420,9 +420,17 @@ int bnxt_qplib_query_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp); int bnxt_qplib_destroy_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp); void *bnxt_qplib_get_qp1_sq_buf(struct bnxt_qplib_qp *qp, struct bnxt_qplib_sge *sge); +void *bnxt_qplib_get_qp1_rq_buf(struct bnxt_qplib_qp *qp, + struct bnxt_qplib_sge *sge); +u32 bnxt_qplib_get_rq_prod_index(struct bnxt_qplib_qp *qp); +dma_addr_t bnxt_qplib_get_qp_buf_from_index(struct bnxt_qplib_qp *qp, + u32 index); void bnxt_qplib_post_send_db(struct bnxt_qplib_qp *qp); int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp, struct bnxt_qplib_swqe *wqe); +void bnxt_qplib_post_recv_db(struct bnxt_qplib_qp *qp); +int bnxt_qplib_post_recv(struct bnxt_qplib_qp *qp, + struct bnxt_qplib_swqe *wqe); int bnxt_qplib_create_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq); int bnxt_qplib_destroy_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq); diff --git a/drivers/infiniband/hw/bnxtre/bnxt_re_ib_verbs.c b/drivers/infiniband/hw/bnxtre/bnxt_re_ib_verbs.c index ff406e9..20a66ec 100644 --- a/drivers/infiniband/hw/bnxtre/bnxt_re_ib_verbs.c +++ b/drivers/infiniband/hw/bnxtre/bnxt_re_ib_verbs.c @@ -1637,6 +1637,51 @@ static int bnxt_re_build_qp1_send_v2(struct bnxt_re_qp *qp, return rc; } +/* For the MAD layer, it only provides the recv SGE the size of + * ib_grh + MAD datagram. No Ethernet headers, Ethertype, BTH, DETH, + * nor RoCE iCRC. The Cu+ solution must provide buffer for the entire + * receive packet (334 bytes) with no VLAN and then copy the GRH + * and the MAD datagram out to the provided SGE. + */ +static int bnxt_re_build_qp1_shadow_qp_recv(struct bnxt_re_qp *qp, + struct ib_recv_wr *wr, + struct bnxt_qplib_swqe *wqe, + int payload_size) +{ + struct bnxt_qplib_sge ref, sge; + u32 rq_prod_index; + struct bnxt_re_sqp_entries *sqp_entry; + + rq_prod_index = bnxt_qplib_get_rq_prod_index(&qp->qplib_qp); + + if (bnxt_qplib_get_qp1_rq_buf(&qp->qplib_qp, &sge)) { + /* Create 1 SGE to receive the entire + * ethernet packet + */ + /* Save the reference from ULP */ + ref.addr = wqe->sg_list[0].addr; + ref.lkey = wqe->sg_list[0].lkey; + ref.size = wqe->sg_list[0].size; + + sqp_entry = &qp->rdev->sqp_tbl[rq_prod_index]; + + /* SGE 1 */ + wqe->sg_list[0].addr = sge.addr; + wqe->sg_list[0].lkey = sge.lkey; + wqe->sg_list[0].size = BNXT_QPLIB_MAX_QP1_RQ_HDR_SIZE_V2; + sge.size -= wqe->sg_list[0].size; + + sqp_entry->sge.addr = ref.addr; + sqp_entry->sge.lkey = ref.lkey; + sqp_entry->sge.size = ref.size; + /* Store the wrid for reporting completion */ + sqp_entry->wrid = wqe->wr_id; + /* change the wqe->wrid to table index */ + wqe->wr_id = rq_prod_index; + } + return 0; +} + static int is_ud_qp(struct bnxt_re_qp *qp) { return qp->qplib_qp.type == CMDQ_CREATE_QP_TYPE_UD; @@ -1983,6 +2028,84 @@ int bnxt_re_post_send(struct ib_qp *ib_qp, struct ib_send_wr *wr, return rc; } +static int bnxt_re_post_recv_shadow_qp(struct bnxt_re_dev *rdev, + struct bnxt_re_qp *qp, + struct ib_recv_wr *wr) +{ + struct bnxt_qplib_swqe wqe; + int rc = 0, payload_sz = 0; + + memset(&wqe, 0, sizeof(wqe)); + while (wr) { + /* House keeping */ + memset(&wqe, 0, sizeof(wqe)); + + /* Common */ + wqe.num_sge = wr->num_sge; + if (wr->num_sge > qp->qplib_qp.rq.max_sge) { + dev_err(rdev_to_dev(rdev), + "Limit exceeded for Receive SGEs"); + rc = -EINVAL; + goto bad; + } + payload_sz = bnxt_re_build_sgl(wr->sg_list, wqe.sg_list, + wr->num_sge); + wqe.wr_id = wr->wr_id; + wqe.type = BNXT_QPLIB_SWQE_TYPE_RECV; + + if (!rc) + rc = bnxt_qplib_post_recv(&qp->qplib_qp, &wqe); +bad: + if (rc) + break; + + wr = wr->next; + } + bnxt_qplib_post_recv_db(&qp->qplib_qp); + return rc; +} + +int bnxt_re_post_recv(struct ib_qp *ib_qp, struct ib_recv_wr *wr, + struct ib_recv_wr **bad_wr) +{ + struct bnxt_re_qp *qp = container_of(ib_qp, struct bnxt_re_qp, ib_qp); + struct bnxt_qplib_swqe wqe; + int rc = 0, payload_sz = 0; + + while (wr) { + /* House keeping */ + memset(&wqe, 0, sizeof(wqe)); + + /* Common */ + wqe.num_sge = wr->num_sge; + if (wr->num_sge > qp->qplib_qp.rq.max_sge) { + dev_err(rdev_to_dev(qp->rdev), + "Limit exceeded for Receive SGEs"); + rc = -EINVAL; + goto bad; + } + + payload_sz = bnxt_re_build_sgl(wr->sg_list, wqe.sg_list, + wr->num_sge); + wqe.wr_id = wr->wr_id; + wqe.type = BNXT_QPLIB_SWQE_TYPE_RECV; + + if (ib_qp->qp_type == IB_QPT_GSI) + rc = bnxt_re_build_qp1_shadow_qp_recv(qp, wr, &wqe, + payload_sz); + if (!rc) + rc = bnxt_qplib_post_recv(&qp->qplib_qp, &wqe); +bad: + if (rc) { + *bad_wr = wr; + break; + } + wr = wr->next; + } + bnxt_qplib_post_recv_db(&qp->qplib_qp); + return rc; +} + /* Completion Queues */ int bnxt_re_destroy_cq(struct ib_cq *ib_cq) { diff --git a/drivers/infiniband/hw/bnxtre/bnxt_re_ib_verbs.h b/drivers/infiniband/hw/bnxtre/bnxt_re_ib_verbs.h index becdcdc..9f3dd49 100644 --- a/drivers/infiniband/hw/bnxtre/bnxt_re_ib_verbs.h +++ b/drivers/infiniband/hw/bnxtre/bnxt_re_ib_verbs.h @@ -164,6 +164,8 @@ int bnxt_re_query_qp(struct ib_qp *qp, struct ib_qp_attr *qp_attr, int bnxt_re_destroy_qp(struct ib_qp *qp); int bnxt_re_post_send(struct ib_qp *qp, struct ib_send_wr *send_wr, struct ib_send_wr **bad_send_wr); +int bnxt_re_post_recv(struct ib_qp *qp, struct ib_recv_wr *recv_wr, + struct ib_recv_wr **bad_recv_wr); struct ib_cq *bnxt_re_create_cq(struct ib_device *ibdev, const struct ib_cq_init_attr *attr, struct ib_ucontext *context, diff --git a/drivers/infiniband/hw/bnxtre/bnxt_re_main.c b/drivers/infiniband/hw/bnxtre/bnxt_re_main.c index 03a34c0..adaa663 100644 --- a/drivers/infiniband/hw/bnxtre/bnxt_re_main.c +++ b/drivers/infiniband/hw/bnxtre/bnxt_re_main.c @@ -463,6 +463,8 @@ static int bnxt_re_register_ib(struct bnxt_re_dev *rdev) ibdev->destroy_qp = bnxt_re_destroy_qp; ibdev->post_send = bnxt_re_post_send; + ibdev->post_recv = bnxt_re_post_recv; + ibdev->create_cq = bnxt_re_create_cq; ibdev->destroy_cq = bnxt_re_destroy_cq; ibdev->req_notify_cq = bnxt_re_req_notify_cq;