Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753131AbeADTCY (ORCPT + 1 other); Thu, 4 Jan 2018 14:02:24 -0500 Received: from ale.deltatee.com ([207.54.116.67]:39290 "EHLO ale.deltatee.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753006AbeADTCJ (ORCPT ); Thu, 4 Jan 2018 14:02:09 -0500 From: Logan Gunthorpe To: linux-kernel@vger.kernel.org, linux-pci@vger.kernel.org, linux-nvme@lists.infradead.org, linux-rdma@vger.kernel.org, linux-nvdimm@lists.01.org, linux-block@vger.kernel.org Cc: Stephen Bates , Christoph Hellwig , Jens Axboe , Keith Busch , Sagi Grimberg , Bjorn Helgaas , Jason Gunthorpe , Max Gurtovoy , Dan Williams , =?UTF-8?q?J=C3=A9r=C3=B4me=20Glisse?= , Benjamin Herrenschmidt , Logan Gunthorpe Date: Thu, 4 Jan 2018 12:01:31 -0700 Message-Id: <20180104190137.7654-7-logang@deltatee.com> X-Mailer: git-send-email 2.11.0 In-Reply-To: <20180104190137.7654-1-logang@deltatee.com> References: <20180104190137.7654-1-logang@deltatee.com> X-SA-Exim-Connect-IP: 172.16.1.31 X-SA-Exim-Rcpt-To: linux-nvme@lists.infradead.org, linux-nvdimm@lists.01.org, linux-kernel@vger.kernel.org, linux-pci@vger.kernel.org, linux-rdma@vger.kernel.org, linux-block@vger.kernel.org, sbates@raithlin.com, hch@lst.de, axboe@kernel.dk, sagi@grimberg.me, bhelgaas@google.com, jgg@mellanox.com, maxg@mellanox.com, keith.busch@intel.com, dan.j.williams@intel.com, jglisse@redhat.com, benh@kernel.crashing.org, logang@deltatee.com X-SA-Exim-Mail-From: gunthorp@deltatee.com Subject: [PATCH 06/12] IB/core: Add optional PCI P2P flag to rdma_rw_ctx_[init|destroy]() X-SA-Exim-Version: 4.2.1 (built Tue, 02 Aug 2016 21:08:31 +0000) X-SA-Exim-Scanned: Yes (on ale.deltatee.com) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Return-Path: In order to use PCI P2P memory pci_p2pmem_[un]map_sg() functions must be called to map the correct DMA address. To do this, we add a flags variable and the RDMA_RW_CTX_FLAG_PCI_P2P flag. When the flag is specified use the appropriate map function. Signed-off-by: Logan Gunthorpe --- drivers/infiniband/core/rw.c | 22 +++++++++++++++++----- drivers/infiniband/ulp/isert/ib_isert.c | 5 +++-- drivers/infiniband/ulp/srpt/ib_srpt.c | 7 ++++--- drivers/nvme/target/rdma.c | 6 +++--- include/rdma/rw.h | 7 +++++-- net/sunrpc/xprtrdma/svc_rdma_rw.c | 6 +++--- 6 files changed, 35 insertions(+), 18 deletions(-) diff --git a/drivers/infiniband/core/rw.c b/drivers/infiniband/core/rw.c index c8963e91f92a..7956484da082 100644 --- a/drivers/infiniband/core/rw.c +++ b/drivers/infiniband/core/rw.c @@ -12,6 +12,7 @@ */ #include #include +#include #include #include @@ -269,18 +270,24 @@ static int rdma_rw_init_single_wr(struct rdma_rw_ctx *ctx, struct ib_qp *qp, * @remote_addr:remote address to read/write (relative to @rkey) * @rkey: remote key to operate on * @dir: %DMA_TO_DEVICE for RDMA WRITE, %DMA_FROM_DEVICE for RDMA READ + * @flags: any of the RDMA_RW_CTX_FLAG_* flags * * Returns the number of WQEs that will be needed on the workqueue if * successful, or a negative error code. */ int rdma_rw_ctx_init(struct rdma_rw_ctx *ctx, struct ib_qp *qp, u8 port_num, struct scatterlist *sg, u32 sg_cnt, u32 sg_offset, - u64 remote_addr, u32 rkey, enum dma_data_direction dir) + u64 remote_addr, u32 rkey, enum dma_data_direction dir, + unsigned int flags) { struct ib_device *dev = qp->pd->device; int ret; - ret = ib_dma_map_sg(dev, sg, sg_cnt, dir); + if (flags & RDMA_RW_CTX_FLAG_PCI_P2P) + ret = pci_p2pmem_map_sg(sg, sg_cnt); + else + ret = ib_dma_map_sg(dev, sg, sg_cnt, dir); + if (!ret) return -ENOMEM; sg_cnt = ret; @@ -499,7 +506,7 @@ struct ib_send_wr *rdma_rw_ctx_wrs(struct rdma_rw_ctx *ctx, struct ib_qp *qp, rdma_rw_update_lkey(&ctx->sig->data, true); if (ctx->sig->prot.mr) rdma_rw_update_lkey(&ctx->sig->prot, true); - + ctx->sig->sig_mr->need_inval = true; ib_update_fast_reg_key(ctx->sig->sig_mr, ib_inc_rkey(ctx->sig->sig_mr->lkey)); @@ -579,9 +586,11 @@ EXPORT_SYMBOL(rdma_rw_ctx_post); * @sg: scatterlist that was used for the READ/WRITE * @sg_cnt: number of entries in @sg * @dir: %DMA_TO_DEVICE for RDMA WRITE, %DMA_FROM_DEVICE for RDMA READ + * @flags: the same flags used to init the context */ void rdma_rw_ctx_destroy(struct rdma_rw_ctx *ctx, struct ib_qp *qp, u8 port_num, - struct scatterlist *sg, u32 sg_cnt, enum dma_data_direction dir) + struct scatterlist *sg, u32 sg_cnt, enum dma_data_direction dir, + unsigned int flags) { int i; @@ -602,7 +611,10 @@ void rdma_rw_ctx_destroy(struct rdma_rw_ctx *ctx, struct ib_qp *qp, u8 port_num, break; } - ib_dma_unmap_sg(qp->pd->device, sg, sg_cnt, dir); + if (flags & RDMA_RW_CTX_FLAG_PCI_P2P) + pci_p2pmem_unmap_sg(sg, sg_cnt); + else + ib_dma_unmap_sg(qp->pd->device, sg, sg_cnt, dir); } EXPORT_SYMBOL(rdma_rw_ctx_destroy); diff --git a/drivers/infiniband/ulp/isert/ib_isert.c b/drivers/infiniband/ulp/isert/ib_isert.c index 720dfb3a1ac2..a076da2ead16 100644 --- a/drivers/infiniband/ulp/isert/ib_isert.c +++ b/drivers/infiniband/ulp/isert/ib_isert.c @@ -1496,7 +1496,8 @@ isert_rdma_rw_ctx_destroy(struct isert_cmd *cmd, struct isert_conn *conn) se_cmd->t_prot_nents, dir); } else { rdma_rw_ctx_destroy(&cmd->rw, conn->qp, conn->cm_id->port_num, - se_cmd->t_data_sg, se_cmd->t_data_nents, dir); + se_cmd->t_data_sg, se_cmd->t_data_nents, + dir, 0); } cmd->rw.nr_ops = 0; @@ -2148,7 +2149,7 @@ isert_rdma_rw_ctx_post(struct isert_cmd *cmd, struct isert_conn *conn, } else { ret = rdma_rw_ctx_init(&cmd->rw, conn->qp, port_num, se_cmd->t_data_sg, se_cmd->t_data_nents, - offset, addr, rkey, dir); + offset, addr, rkey, dir, 0); } if (ret < 0) { isert_err("Cmd: %p failed to prepare RDMA res\n", cmd); diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c index 8a1bd354b1cc..c5371ab2e47d 100644 --- a/drivers/infiniband/ulp/srpt/ib_srpt.c +++ b/drivers/infiniband/ulp/srpt/ib_srpt.c @@ -854,7 +854,8 @@ static int srpt_alloc_rw_ctxs(struct srpt_send_ioctx *ioctx, goto unwind; ret = rdma_rw_ctx_init(&ctx->rw, ch->qp, ch->sport->port, - ctx->sg, ctx->nents, 0, remote_addr, rkey, dir); + ctx->sg, ctx->nents, 0, remote_addr, rkey, + dir, 0); if (ret < 0) { target_free_sgl(ctx->sg, ctx->nents); goto unwind; @@ -883,7 +884,7 @@ static int srpt_alloc_rw_ctxs(struct srpt_send_ioctx *ioctx, struct srpt_rw_ctx *ctx = &ioctx->rw_ctxs[i]; rdma_rw_ctx_destroy(&ctx->rw, ch->qp, ch->sport->port, - ctx->sg, ctx->nents, dir); + ctx->sg, ctx->nents, dir, 0); target_free_sgl(ctx->sg, ctx->nents); } if (ioctx->rw_ctxs != &ioctx->s_rw_ctx) @@ -901,7 +902,7 @@ static void srpt_free_rw_ctxs(struct srpt_rdma_ch *ch, struct srpt_rw_ctx *ctx = &ioctx->rw_ctxs[i]; rdma_rw_ctx_destroy(&ctx->rw, ch->qp, ch->sport->port, - ctx->sg, ctx->nents, dir); + ctx->sg, ctx->nents, dir, 0); target_free_sgl(ctx->sg, ctx->nents); } diff --git a/drivers/nvme/target/rdma.c b/drivers/nvme/target/rdma.c index 49912909c298..d4d0662ab071 100644 --- a/drivers/nvme/target/rdma.c +++ b/drivers/nvme/target/rdma.c @@ -480,7 +480,7 @@ static void nvmet_rdma_release_rsp(struct nvmet_rdma_rsp *rsp) if (rsp->n_rdma) { rdma_rw_ctx_destroy(&rsp->rw, queue->cm_id->qp, queue->cm_id->port_num, rsp->req.sg, - rsp->req.sg_cnt, nvmet_data_dir(&rsp->req)); + rsp->req.sg_cnt, nvmet_data_dir(&rsp->req), 0); } if (rsp->req.sg != &rsp->cmd->inline_sg) @@ -563,7 +563,7 @@ static void nvmet_rdma_read_data_done(struct ib_cq *cq, struct ib_wc *wc) atomic_add(rsp->n_rdma, &queue->sq_wr_avail); rdma_rw_ctx_destroy(&rsp->rw, queue->cm_id->qp, queue->cm_id->port_num, rsp->req.sg, - rsp->req.sg_cnt, nvmet_data_dir(&rsp->req)); + rsp->req.sg_cnt, nvmet_data_dir(&rsp->req), 0); rsp->n_rdma = 0; if (unlikely(wc->status != IB_WC_SUCCESS)) { @@ -634,7 +634,7 @@ static u16 nvmet_rdma_map_sgl_keyed(struct nvmet_rdma_rsp *rsp, ret = rdma_rw_ctx_init(&rsp->rw, cm_id->qp, cm_id->port_num, rsp->req.sg, rsp->req.sg_cnt, 0, addr, key, - nvmet_data_dir(&rsp->req)); + nvmet_data_dir(&rsp->req), 0); if (ret < 0) return NVME_SC_INTERNAL; rsp->req.transfer_len += len; diff --git a/include/rdma/rw.h b/include/rdma/rw.h index a3cbbc7b6417..ba8050434667 100644 --- a/include/rdma/rw.h +++ b/include/rdma/rw.h @@ -59,12 +59,15 @@ struct rdma_rw_ctx { }; }; +#define RDMA_RW_CTX_FLAG_PCI_P2P (1 << 0) + int rdma_rw_ctx_init(struct rdma_rw_ctx *ctx, struct ib_qp *qp, u8 port_num, struct scatterlist *sg, u32 sg_cnt, u32 sg_offset, - u64 remote_addr, u32 rkey, enum dma_data_direction dir); + u64 remote_addr, u32 rkey, enum dma_data_direction dir, + unsigned int flags); void rdma_rw_ctx_destroy(struct rdma_rw_ctx *ctx, struct ib_qp *qp, u8 port_num, struct scatterlist *sg, u32 sg_cnt, - enum dma_data_direction dir); + enum dma_data_direction dir, unsigned int flags); int rdma_rw_ctx_signature_init(struct rdma_rw_ctx *ctx, struct ib_qp *qp, u8 port_num, struct scatterlist *sg, u32 sg_cnt, diff --git a/net/sunrpc/xprtrdma/svc_rdma_rw.c b/net/sunrpc/xprtrdma/svc_rdma_rw.c index 9bd04549a1ad..5f46c35e6707 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_rw.c +++ b/net/sunrpc/xprtrdma/svc_rdma_rw.c @@ -140,7 +140,7 @@ static void svc_rdma_cc_release(struct svc_rdma_chunk_ctxt *cc, rdma_rw_ctx_destroy(&ctxt->rw_ctx, rdma->sc_qp, rdma->sc_port_num, ctxt->rw_sg_table.sgl, - ctxt->rw_nents, dir); + ctxt->rw_nents, dir, 0); svc_rdma_put_rw_ctxt(rdma, ctxt); } svc_xprt_put(&rdma->sc_xprt); @@ -433,7 +433,7 @@ svc_rdma_build_writes(struct svc_rdma_write_info *info, ret = rdma_rw_ctx_init(&ctxt->rw_ctx, rdma->sc_qp, rdma->sc_port_num, ctxt->rw_sg_table.sgl, ctxt->rw_nents, 0, seg_offset, - seg_handle, DMA_TO_DEVICE); + seg_handle, DMA_TO_DEVICE, 0); if (ret < 0) goto out_initerr; @@ -639,7 +639,7 @@ static int svc_rdma_build_read_segment(struct svc_rdma_read_info *info, ret = rdma_rw_ctx_init(&ctxt->rw_ctx, cc->cc_rdma->sc_qp, cc->cc_rdma->sc_port_num, ctxt->rw_sg_table.sgl, ctxt->rw_nents, - 0, offset, rkey, DMA_FROM_DEVICE); + 0, offset, rkey, DMA_FROM_DEVICE, 0); if (ret < 0) goto out_initerr; -- 2.11.0