| Message ID | 1499691002-5224-1-git-send-email-sagi@grimberg.me (mailing list archive) |
|---|---|
| State | RFC |
| Headers | show |
> On Jul 10, 2017, at 8:50 AM, Sagi Grimberg <sagi@grimberg.me> wrote: > > Currently FMRs don't even expose maximum supported page list length, > which makes it impossible for to know what it actually is. I thought the FMR page list limit was always 64. I seem to vaguely recall testing this way back when I split out the xprtrdma code for handling different registration modes, and above 64, FMR fell over (at least with CX2 and CX3). I don't have a problem with limiting FMR to 64 pages. It would be nice to have a core-defined constant for that, though. > Thus, rename max_fast_reg_page_list_len to max_reg_page_list_len > so it is the same capability regardless of registration method is > used. Drivers that support both methods should expose the minimum > between the two. I would hate to limit the FRWR page depth to 64. What's the rationale behind making these the same? One other comment below. > Signed-off-by: Sagi Grimberg <sagi@grimberg.me> > --- > drivers/infiniband/core/rw.c | 2 +- > drivers/infiniband/hw/bnxt_re/ib_verbs.c | 2 +- > drivers/infiniband/hw/cxgb3/iwch_provider.c | 2 +- > drivers/infiniband/hw/cxgb4/provider.c | 2 +- > drivers/infiniband/hw/hfi1/verbs.c | 2 +- > drivers/infiniband/hw/i40iw/i40iw_verbs.c | 2 +- > drivers/infiniband/hw/mlx4/main.c | 2 +- > drivers/infiniband/hw/mlx5/main.c | 2 +- > drivers/infiniband/hw/mthca/mthca_provider.c | 5 +++++ > drivers/infiniband/hw/ocrdma/ocrdma_verbs.c | 2 +- > drivers/infiniband/hw/qedr/verbs.c | 2 +- > drivers/infiniband/hw/qib/qib_verbs.c | 1 + > drivers/infiniband/hw/usnic/usnic_ib_verbs.c | 2 +- > drivers/infiniband/sw/rxe/rxe.c | 2 +- > drivers/infiniband/ulp/iser/iser_verbs.c | 2 +- > drivers/infiniband/ulp/srp/ib_srp.c | 21 +++------------------ > drivers/infiniband/ulp/srp/ib_srp.h | 2 -- > drivers/nvme/host/rdma.c | 2 +- > include/rdma/ib_verbs.h | 2 +- > net/sunrpc/xprtrdma/fmr_ops.c | 10 ++++++++-- > net/sunrpc/xprtrdma/frwr_ops.c | 4 ++-- > net/sunrpc/xprtrdma/svc_rdma_transport.c | 2 +- > 22 files changed, 35 insertions(+), 40 deletions(-) > > diff --git a/drivers/infiniband/core/rw.c b/drivers/infiniband/core/rw.c > index dbfd854c32c9..cd4adb24a077 100644 > --- a/drivers/infiniband/core/rw.c > +++ b/drivers/infiniband/core/rw.c > @@ -61,7 +61,7 @@ static inline bool rdma_rw_io_needs_mr(struct ib_device *dev, u8 port_num, > static inline u32 rdma_rw_fr_page_list_len(struct ib_device *dev) > { > /* arbitrary limit to avoid allocating gigantic resources */ > - return min_t(u32, dev->attrs.max_fast_reg_page_list_len, 256); > + return min_t(u32, dev->attrs.max_reg_page_list_len, 256); > } > > /* Caller must have zero-initialized *reg. */ > diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c > index c7bd68311d0c..2c9a8eed6183 100644 > --- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c > +++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c > @@ -198,7 +198,7 @@ int bnxt_re_query_device(struct ib_device *ibdev, > ib_attr->max_srq_wr = dev_attr->max_srq_wqes; > ib_attr->max_srq_sge = dev_attr->max_srq_sges; > > - ib_attr->max_fast_reg_page_list_len = MAX_PBL_LVL_1_PGS; > + ib_attr->max_reg_page_list_len = MAX_PBL_LVL_1_PGS; > > ib_attr->max_pkeys = 1; > ib_attr->local_ca_ack_delay = 0; > diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c > index 29d30744d6c9..bcda349eaefd 100644 > --- a/drivers/infiniband/hw/cxgb3/iwch_provider.c > +++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c > @@ -1112,7 +1112,7 @@ static int iwch_query_device(struct ib_device *ibdev, struct ib_device_attr *pro > props->max_mr = dev->attr.max_mem_regs; > props->max_pd = dev->attr.max_pds; > props->local_ca_ack_delay = 0; > - props->max_fast_reg_page_list_len = T3_MAX_FASTREG_DEPTH; > + props->max_reg_page_list_len = T3_MAX_FASTREG_DEPTH; > > return 0; > } > diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c > index 0771e9a4d061..7d2f9136deb7 100644 > --- a/drivers/infiniband/hw/cxgb4/provider.c > +++ b/drivers/infiniband/hw/cxgb4/provider.c > @@ -351,7 +351,7 @@ static int c4iw_query_device(struct ib_device *ibdev, struct ib_device_attr *pro > props->max_mr = c4iw_num_stags(&dev->rdev); > props->max_pd = T4_MAX_NUM_PD; > props->local_ca_ack_delay = 0; > - props->max_fast_reg_page_list_len = > + props->max_reg_page_list_len = > t4_max_fr_depth(dev->rdev.lldi.ulptx_memwrite_dsgl && use_dsgl); > > return 0; > diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c > index 5b53faf47042..2b15ee143e45 100644 > --- a/drivers/infiniband/hw/hfi1/verbs.c > +++ b/drivers/infiniband/hw/hfi1/verbs.c > @@ -1284,7 +1284,7 @@ static void hfi1_fill_device_attr(struct hfi1_devdata *dd) > rdi->dparms.props.hw_ver = dd->minrev; > rdi->dparms.props.sys_image_guid = ib_hfi1_sys_image_guid; > rdi->dparms.props.max_mr_size = U64_MAX; > - rdi->dparms.props.max_fast_reg_page_list_len = UINT_MAX; > + rdi->dparms.props.max_reg_page_list_len = UINT_MAX; > rdi->dparms.props.max_qp = hfi1_max_qps; > rdi->dparms.props.max_qp_wr = hfi1_max_qp_wrs; > rdi->dparms.props.max_sge = hfi1_max_sges; > diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c > index 4dbe61ec7a77..77bc44ae7ce6 100644 > --- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c > +++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c > @@ -80,7 +80,7 @@ static int i40iw_query_device(struct ib_device *ibdev, > props->max_qp_init_rd_atom = props->max_qp_rd_atom; > props->atomic_cap = IB_ATOMIC_NONE; > props->max_map_per_fmr = 1; > - props->max_fast_reg_page_list_len = I40IW_MAX_PAGES_PER_FMR; > + props->max_reg_page_list_len = I40IW_MAX_PAGES_PER_FMR; > return 0; > } > > diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c > index 75b2f7d4cd95..10e203a2fd4c 100644 > --- a/drivers/infiniband/hw/mlx4/main.c > +++ b/drivers/infiniband/hw/mlx4/main.c > @@ -537,7 +537,7 @@ static int mlx4_ib_query_device(struct ib_device *ibdev, > props->max_srq = dev->dev->quotas.srq; > props->max_srq_wr = dev->dev->caps.max_srq_wqes - 1; > props->max_srq_sge = dev->dev->caps.max_srq_sge; > - props->max_fast_reg_page_list_len = MLX4_MAX_FAST_REG_PAGES; > + props->max_reg_page_list_len = MLX4_MAX_FAST_REG_PAGES; > props->local_ca_ack_delay = dev->dev->caps.local_ca_ack_delay; > props->atomic_cap = dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_ATOMIC ? > IB_ATOMIC_HCA : IB_ATOMIC_NONE; > diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c > index 2a55372b7da2..0452c9f9d432 100644 > --- a/drivers/infiniband/hw/mlx5/main.c > +++ b/drivers/infiniband/hw/mlx5/main.c > @@ -728,7 +728,7 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, > props->local_ca_ack_delay = MLX5_CAP_GEN(mdev, local_ca_ack_delay); > props->max_res_rd_atom = props->max_qp_rd_atom * props->max_qp; > props->max_srq_sge = max_rq_sg - 1; > - props->max_fast_reg_page_list_len = > + props->max_reg_page_list_len = > 1 << MLX5_CAP_GEN(mdev, log_max_klm_list_size); > get_atomic_caps(dev, props); > props->masked_atomic_cap = IB_ATOMIC_NONE; > diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c > index c197cd9b193f..93a7a72f7c36 100644 > --- a/drivers/infiniband/hw/mthca/mthca_provider.c > +++ b/drivers/infiniband/hw/mthca/mthca_provider.c > @@ -116,6 +116,11 @@ static int mthca_query_device(struct ib_device *ibdev, struct ib_device_attr *pr > props->max_mcast_qp_attach = MTHCA_QP_PER_MGM; > props->max_total_mcast_qp_attach = props->max_mcast_qp_attach * > props->max_mcast_grp; > + if (mthca_is_memfree(mdev)) > + /* For Arbel, all MTTs must fit in the same page. */ > + props->max_reg_page_list_len = PAGE_SIZE / sizeof(u64); > + else > + props->max_reg_page_list_len = INT_MAX; > /* > * If Sinai memory key optimization is being used, then only > * the 8-bit key portion will change. For other HCAs, the > diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c > index 2f30bda8457a..d0ac9af99ac7 100644 > --- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c > +++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c > @@ -142,7 +142,7 @@ int ocrdma_query_device(struct ib_device *ibdev, struct ib_device_attr *attr, > attr->max_srq_sge = dev->attr.max_srq_sge; > attr->max_srq_wr = dev->attr.max_rqe; > attr->local_ca_ack_delay = dev->attr.local_ca_ack_delay; > - attr->max_fast_reg_page_list_len = dev->attr.max_pages_per_frmr; > + attr->max_reg_page_list_len = dev->attr.max_pages_per_frmr; > attr->max_pkeys = 1; > return 0; > } > diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c > index d6723c365c7f..35691df9cdf2 100644 > --- a/drivers/infiniband/hw/qedr/verbs.c > +++ b/drivers/infiniband/hw/qedr/verbs.c > @@ -163,7 +163,7 @@ int qedr_query_device(struct ib_device *ibdev, > attr->max_srq_wr = qattr->max_srq_wr; > > attr->local_ca_ack_delay = qattr->dev_ack_delay; > - attr->max_fast_reg_page_list_len = qattr->max_mr / 8; > + attr->max_reg_page_list_len = qattr->max_mr / 8; > attr->max_pkeys = QEDR_ROCE_PKEY_MAX; > attr->max_ah = qattr->max_ah; > > diff --git a/drivers/infiniband/hw/qib/qib_verbs.c b/drivers/infiniband/hw/qib/qib_verbs.c > index ac42dce7e281..ff92c44083f7 100644 > --- a/drivers/infiniband/hw/qib/qib_verbs.c > +++ b/drivers/infiniband/hw/qib/qib_verbs.c > @@ -1501,6 +1501,7 @@ static void qib_fill_device_attr(struct qib_devdata *dd) > rdi->dparms.props.max_total_mcast_qp_attach = > rdi->dparms.props.max_mcast_qp_attach * > rdi->dparms.props.max_mcast_grp; > + rdi->dparms.props.max_reg_page_list_len = UINT_MAX; > /* post send table */ > dd->verbs_dev.rdi.post_parms = qib_post_parms; > } > diff --git a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c > index 4996984885c2..a173de1761ca 100644 > --- a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c > +++ b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c > @@ -310,7 +310,7 @@ int usnic_ib_query_device(struct ib_device *ibdev, > props->max_srq = 0; > props->max_srq_wr = 0; > props->max_srq_sge = 0; > - props->max_fast_reg_page_list_len = 0; > + props->max_reg_page_list_len = 0; > props->max_mcast_grp = 0; > props->max_mcast_qp_attach = 0; > props->max_total_mcast_qp_attach = 0; > diff --git a/drivers/infiniband/sw/rxe/rxe.c b/drivers/infiniband/sw/rxe/rxe.c > index c21c913f911a..e30fd9694902 100644 > --- a/drivers/infiniband/sw/rxe/rxe.c > +++ b/drivers/infiniband/sw/rxe/rxe.c > @@ -124,7 +124,7 @@ static int rxe_init_device_param(struct rxe_dev *rxe) > rxe->attr.max_srq = RXE_MAX_SRQ; > rxe->attr.max_srq_wr = RXE_MAX_SRQ_WR; > rxe->attr.max_srq_sge = RXE_MAX_SRQ_SGE; > - rxe->attr.max_fast_reg_page_list_len = RXE_MAX_FMR_PAGE_LIST_LEN; > + rxe->attr.max_reg_page_list_len = RXE_MAX_FMR_PAGE_LIST_LEN; > rxe->attr.max_pkeys = RXE_MAX_PKEYS; > rxe->attr.local_ca_ack_delay = RXE_LOCAL_CA_ACK_DELAY; > > diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c > index c538a38c91ce..28f26ab4585d 100644 > --- a/drivers/infiniband/ulp/iser/iser_verbs.c > +++ b/drivers/infiniband/ulp/iser/iser_verbs.c > @@ -709,7 +709,7 @@ iser_calc_scsi_params(struct iser_conn *iser_conn, > > sg_tablesize = DIV_ROUND_UP(max_sectors * 512, SIZE_4K); > sup_sg_tablesize = min_t(unsigned, ISCSI_ISER_MAX_SG_TABLESIZE, > - device->ib_device->attrs.max_fast_reg_page_list_len); > + device->ib_device->attrs.max_reg_page_list_len); > > iser_conn->scsi_sg_tablesize = min(sg_tablesize, sup_sg_tablesize); > } > diff --git a/drivers/infiniband/ulp/srp/ib_srp.c b/drivers/infiniband/ulp/srp/ib_srp.c > index 2354c742caa1..d752100286f9 100644 > --- a/drivers/infiniband/ulp/srp/ib_srp.c > +++ b/drivers/infiniband/ulp/srp/ib_srp.c > @@ -3580,7 +3580,6 @@ static void srp_add_one(struct ib_device *device) > struct ib_device_attr *attr = &device->attrs; > struct srp_host *host; > int mr_page_shift, p; > - u64 max_pages_per_mr; > unsigned int flags = 0; > > srp_dev = kzalloc(sizeof(*srp_dev), GFP_KERNEL); > @@ -3595,13 +3594,8 @@ static void srp_add_one(struct ib_device *device) > mr_page_shift = max(12, ffs(attr->page_size_cap) - 1); > srp_dev->mr_page_size = 1 << mr_page_shift; > srp_dev->mr_page_mask = ~((u64) srp_dev->mr_page_size - 1); > - max_pages_per_mr = attr->max_mr_size; > - do_div(max_pages_per_mr, srp_dev->mr_page_size); > - pr_debug("%s: %llu / %u = %llu <> %u\n", __func__, > - attr->max_mr_size, srp_dev->mr_page_size, > - max_pages_per_mr, SRP_MAX_PAGES_PER_MR); > srp_dev->max_pages_per_mr = min_t(u64, SRP_MAX_PAGES_PER_MR, > - max_pages_per_mr); > + attr->max_reg_page_list_len); > > srp_dev->has_fmr = (device->alloc_fmr && device->dealloc_fmr && > device->map_phys_fmr && device->unmap_fmr); > @@ -3620,17 +3614,8 @@ static void srp_add_one(struct ib_device *device) > (!srp_dev->has_fmr && !srp_dev->has_fr)) > flags |= IB_PD_UNSAFE_GLOBAL_RKEY; > > - if (srp_dev->use_fast_reg) { > - srp_dev->max_pages_per_mr = > - min_t(u32, srp_dev->max_pages_per_mr, > - attr->max_fast_reg_page_list_len); > - } > - srp_dev->mr_max_size = srp_dev->mr_page_size * > - srp_dev->max_pages_per_mr; > - pr_debug("%s: mr_page_shift = %d, device->max_mr_size = %#llx, device->max_fast_reg_page_list_len = %u, max_pages_per_mr = %d, mr_max_size = %#x\n", > - device->name, mr_page_shift, attr->max_mr_size, > - attr->max_fast_reg_page_list_len, > - srp_dev->max_pages_per_mr, srp_dev->mr_max_size); > + pr_debug("%s: mr_page_shift = %d, max_pages_per_mr = %d\n", > + device->name, mr_page_shift, srp_dev->max_pages_per_mr); > > INIT_LIST_HEAD(&srp_dev->dev_list); > > diff --git a/drivers/infiniband/ulp/srp/ib_srp.h b/drivers/infiniband/ulp/srp/ib_srp.h > index ab9077b81d5a..9f2ada3855a5 100644 > --- a/drivers/infiniband/ulp/srp/ib_srp.h > +++ b/drivers/infiniband/ulp/srp/ib_srp.h > @@ -83,7 +83,6 @@ enum srp_iu_type { > /* > * @mr_page_mask: HCA memory registration page mask. > * @mr_page_size: HCA memory registration page size. > - * @mr_max_size: Maximum size in bytes of a single FMR / FR registration > * request. > */ > struct srp_device { > @@ -92,7 +91,6 @@ struct srp_device { > struct ib_pd *pd; > u64 mr_page_mask; > int mr_page_size; > - int mr_max_size; > int max_pages_per_mr; > bool has_fmr; > bool has_fr; > diff --git a/drivers/nvme/host/rdma.c b/drivers/nvme/host/rdma.c > index 24397d306d53..22882783bf71 100644 > --- a/drivers/nvme/host/rdma.c > +++ b/drivers/nvme/host/rdma.c > @@ -1586,7 +1586,7 @@ static int nvme_rdma_configure_admin_queue(struct nvme_rdma_ctrl *ctrl) > goto out_free_queue; > > ctrl->max_fr_pages = min_t(u32, NVME_RDMA_MAX_SEGMENTS, > - ctrl->device->dev->attrs.max_fast_reg_page_list_len); > + ctrl->device->dev->attrs.max_reg_page_list_len); > > memset(&ctrl->admin_tag_set, 0, sizeof(ctrl->admin_tag_set)); > ctrl->admin_tag_set.ops = &nvme_rdma_admin_mq_ops; > diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h > index 9d4d2a74c95e..91359e5c3961 100644 > --- a/include/rdma/ib_verbs.h > +++ b/include/rdma/ib_verbs.h > @@ -327,7 +327,7 @@ struct ib_device_attr { > int max_srq; > int max_srq_wr; > int max_srq_sge; > - unsigned int max_fast_reg_page_list_len; > + unsigned int max_reg_page_list_len; > u16 max_pkeys; > u8 local_ca_ack_delay; > int sig_prot_cap; > diff --git a/net/sunrpc/xprtrdma/fmr_ops.c b/net/sunrpc/xprtrdma/fmr_ops.c > index 59e64025ed96..6000302eede7 100644 > --- a/net/sunrpc/xprtrdma/fmr_ops.c > +++ b/net/sunrpc/xprtrdma/fmr_ops.c > @@ -160,8 +160,12 @@ static int > fmr_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep, > struct rpcrdma_create_data_internal *cdata) > { > + struct ib_device_attr *attrs = &ia->ri_device->attrs; > + > + ia->ri_max_frmr_depth = min_t(unsigned int, RPCRDMA_MAX_DATA_SEGS, > + attrs->max_reg_page_list_len); > ia->ri_max_segs = max_t(unsigned int, 1, RPCRDMA_MAX_DATA_SEGS / > - RPCRDMA_MAX_FMR_SGES); > + ia->ri_max_frmr_depth); I would rename ri_max_frmr_depth if it is going to be used by both registration mode. How about ri_max_mr_page_depth ? > return 0; > } > > @@ -170,8 +174,10 @@ fmr_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep, > static size_t > fmr_op_maxpages(struct rpcrdma_xprt *r_xprt) > { > + struct rpcrdma_ia *ia = &r_xprt->rx_ia; > + > return min_t(unsigned int, RPCRDMA_MAX_DATA_SEGS, > - RPCRDMA_MAX_HDR_SEGS * RPCRDMA_MAX_FMR_SGES); > + RPCRDMA_MAX_HDR_SEGS * ia->ri_max_frmr_depth); > } > > /* Use the ib_map_phys_fmr() verb to register a memory region > diff --git a/net/sunrpc/xprtrdma/frwr_ops.c b/net/sunrpc/xprtrdma/frwr_ops.c > index f81dd93176c0..11effcca0519 100644 > --- a/net/sunrpc/xprtrdma/frwr_ops.c > +++ b/net/sunrpc/xprtrdma/frwr_ops.c > @@ -84,7 +84,7 @@ frwr_is_supported(struct rpcrdma_ia *ia) > > if (!(attrs->device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS)) > goto out_not_supported; > - if (attrs->max_fast_reg_page_list_len == 0) > + if (attrs->max_reg_page_list_len == 0) > goto out_not_supported; > return true; > > @@ -215,7 +215,7 @@ frwr_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep, > > ia->ri_max_frmr_depth = > min_t(unsigned int, RPCRDMA_MAX_DATA_SEGS, > - attrs->max_fast_reg_page_list_len); > + attrs->max_reg_page_list_len); > dprintk("RPC: %s: device's max FR page list len = %u\n", > __func__, ia->ri_max_frmr_depth); > > diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c > index a9d9cb1ba4c6..5725abf739ee 100644 > --- a/net/sunrpc/xprtrdma/svc_rdma_transport.c > +++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c > @@ -1001,7 +1001,7 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt) > newxprt->sc_reader = rdma_read_chunk_lcl; > if (dev->attrs.device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS) { > newxprt->sc_frmr_pg_list_len = > - dev->attrs.max_fast_reg_page_list_len; > + dev->attrs.max_reg_page_list_len; > newxprt->sc_dev_caps |= SVCRDMA_DEVCAP_FAST_REG; > newxprt->sc_reader = rdma_read_chunk_frmr; > } else > -- > 2.7.4 > -- Chuck Lever -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
> I thought the FMR page list limit was always 64. I seem to > vaguely recall testing this way back when I split out the > xprtrdma code for handling different registration modes, > and above 64, FMR fell over (at least with CX2 and CX3). Not sure about CX2, but iser runs fmrs with 128 entries perfectly fine on CX3. IIRC CX2 does not support FRWRs, CX4 and up does not support FMRs (which I'd love to kill altogether anyways). The motivation is to not guess empirically what is the max page_list for FMRs which is supposed to be on its way out anyhow. -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
> On Jul 10, 2017, at 10:26 AM, Sagi Grimberg <sagi@grimberg.me> wrote: > > >> I thought the FMR page list limit was always 64. I seem to >> vaguely recall testing this way back when I split out the >> xprtrdma code for handling different registration modes, >> and above 64, FMR fell over (at least with CX2 and CX3). > > Not sure about CX2, but iser runs fmrs with 128 entries perfectly fine > on CX3. IIRC CX2 does not support FRWRs, CX2 does indeed support both FMR and FRWR. > CX4 and up does not support FMRs (which I'd love to kill altogether > anyways). "Killing in-kernel FMR" might be a good topic for the LPC RDMA micro-conference. > The motivation is to not guess empirically what is the max page_list > for FMRs which is supposed to be on its way out anyhow. 1. I concur that it would be good to have some API for advertising the limit for FMR. If it's higher than 64 for some devices, that will be good for performance. 2. But I'm not sure it's always the same as the device's FRWR maximum page list depth. If it is always or often smaller, then that unfairly limits FRWR. 3. I'm not at all objecting, but if FMR is to go away, why make this change now? -- Chuck Lever -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
>>> I thought the FMR page list limit was always 64. I seem to >>> vaguely recall testing this way back when I split out the >>> xprtrdma code for handling different registration modes, >>> and above 64, FMR fell over (at least with CX2 and CX3). >> >> Not sure about CX2, but iser runs fmrs with 128 entries perfectly fine >> on CX3. IIRC CX2 does not support FRWRs, > > CX2 does indeed support both FMR and FRWR. Thanks for the info. >> CX4 and up does not support FMRs (which I'd love to kill altogether >> anyways). > > "Killing in-kernel FMR" might be a good topic for the LPC RDMA > micro-conference. I won't be able to attend, but I'd love if others can discuss it. I'd like for that to happen, but I'd also like to support as much non-deprecated devices as possible. In fact, until the FMR bug report came along, I had a "remove fmr" patch for iser in the pipe and held it back as someone out there is using iser over qib. So as much as I'd like that to happen, so we can just ignore it, I'm convinced we can do it just yet. > 1. I concur that it would be good to have some API for advertising > the limit for FMR. If it's higher than 64 for some devices, that > will be good for performance. That is what I'm trying to achieve. > 2. But I'm not sure it's always the same as the device's FRWR > maximum page list depth. If it is always or often smaller, then > that unfairly limits FRWR. The opposite argument can be made too. is a separate max_fmr_page_list_len a better way to go? > 3. I'm not at all objecting, but if FMR is to go away, why make > this change now? As I said, I don't know if FMRs are going away, its not on me to decide. I just tried to help ULPs to not guess the limit. -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
> On Jul 10, 2017, at 3:13 PM, Sagi Grimberg <sagi@grimberg.me> wrote: > > >>>> I thought the FMR page list limit was always 64. I seem to >>>> vaguely recall testing this way back when I split out the >>>> xprtrdma code for handling different registration modes, >>>> and above 64, FMR fell over (at least with CX2 and CX3). >>> >>> Not sure about CX2, but iser runs fmrs with 128 entries perfectly fine >>> on CX3. IIRC CX2 does not support FRWRs, >> CX2 does indeed support both FMR and FRWR. > > Thanks for the info. > >>> CX4 and up does not support FMRs (which I'd love to kill altogether >>> anyways). >> "Killing in-kernel FMR" might be a good topic for the LPC RDMA >> micro-conference. > > I won't be able to attend, but I'd love if others can discuss it. > I'd like for that to happen, but I'd also like to support as much > non-deprecated devices as possible. In fact, until the FMR bug report > came along, I had a "remove fmr" patch for iser in the pipe and held > it back as someone out there is using iser over qib. > > So as much as I'd like that to happen, so we can just ignore it, I'm > convinced we can do it just yet. >> 1. I concur that it would be good to have some API for advertising >> the limit for FMR. If it's higher than 64 for some devices, that >> will be good for performance. > > That is what I'm trying to achieve. Thumbs up for that. >> 2. But I'm not sure it's always the same as the device's FRWR >> maximum page list depth. If it is always or often smaller, then >> that unfairly limits FRWR. > > The opposite argument can be made too. > > is a separate max_fmr_page_list_len a better way to go? I would prefer that these be exposed as separate per-device limits, yes, since I'll bet they are going to be significantly different values in some cases. >> 3. I'm not at all objecting, but if FMR is to go away, why make >> this change now? > > As I said, I don't know if FMRs are going away, its not on me to > decide. I just tried to help ULPs to not guess the limit. Fair enough. -- Chuck Lever -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/infiniband/core/rw.c b/drivers/infiniband/core/rw.c index dbfd854c32c9..cd4adb24a077 100644 --- a/drivers/infiniband/core/rw.c +++ b/drivers/infiniband/core/rw.c @@ -61,7 +61,7 @@ static inline bool rdma_rw_io_needs_mr(struct ib_device *dev, u8 port_num, static inline u32 rdma_rw_fr_page_list_len(struct ib_device *dev) { /* arbitrary limit to avoid allocating gigantic resources */ - return min_t(u32, dev->attrs.max_fast_reg_page_list_len, 256); + return min_t(u32, dev->attrs.max_reg_page_list_len, 256); } /* Caller must have zero-initialized *reg. */ diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c index c7bd68311d0c..2c9a8eed6183 100644 --- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c +++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c @@ -198,7 +198,7 @@ int bnxt_re_query_device(struct ib_device *ibdev, ib_attr->max_srq_wr = dev_attr->max_srq_wqes; ib_attr->max_srq_sge = dev_attr->max_srq_sges; - ib_attr->max_fast_reg_page_list_len = MAX_PBL_LVL_1_PGS; + ib_attr->max_reg_page_list_len = MAX_PBL_LVL_1_PGS; ib_attr->max_pkeys = 1; ib_attr->local_ca_ack_delay = 0; diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c index 29d30744d6c9..bcda349eaefd 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_provider.c +++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c @@ -1112,7 +1112,7 @@ static int iwch_query_device(struct ib_device *ibdev, struct ib_device_attr *pro props->max_mr = dev->attr.max_mem_regs; props->max_pd = dev->attr.max_pds; props->local_ca_ack_delay = 0; - props->max_fast_reg_page_list_len = T3_MAX_FASTREG_DEPTH; + props->max_reg_page_list_len = T3_MAX_FASTREG_DEPTH; return 0; } diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c index 0771e9a4d061..7d2f9136deb7 100644 --- a/drivers/infiniband/hw/cxgb4/provider.c +++ b/drivers/infiniband/hw/cxgb4/provider.c @@ -351,7 +351,7 @@ static int c4iw_query_device(struct ib_device *ibdev, struct ib_device_attr *pro props->max_mr = c4iw_num_stags(&dev->rdev); props->max_pd = T4_MAX_NUM_PD; props->local_ca_ack_delay = 0; - props->max_fast_reg_page_list_len = + props->max_reg_page_list_len = t4_max_fr_depth(dev->rdev.lldi.ulptx_memwrite_dsgl && use_dsgl); return 0; diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c index 5b53faf47042..2b15ee143e45 100644 --- a/drivers/infiniband/hw/hfi1/verbs.c +++ b/drivers/infiniband/hw/hfi1/verbs.c @@ -1284,7 +1284,7 @@ static void hfi1_fill_device_attr(struct hfi1_devdata *dd) rdi->dparms.props.hw_ver = dd->minrev; rdi->dparms.props.sys_image_guid = ib_hfi1_sys_image_guid; rdi->dparms.props.max_mr_size = U64_MAX; - rdi->dparms.props.max_fast_reg_page_list_len = UINT_MAX; + rdi->dparms.props.max_reg_page_list_len = UINT_MAX; rdi->dparms.props.max_qp = hfi1_max_qps; rdi->dparms.props.max_qp_wr = hfi1_max_qp_wrs; rdi->dparms.props.max_sge = hfi1_max_sges; diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c index 4dbe61ec7a77..77bc44ae7ce6 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c +++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c @@ -80,7 +80,7 @@ static int i40iw_query_device(struct ib_device *ibdev, props->max_qp_init_rd_atom = props->max_qp_rd_atom; props->atomic_cap = IB_ATOMIC_NONE; props->max_map_per_fmr = 1; - props->max_fast_reg_page_list_len = I40IW_MAX_PAGES_PER_FMR; + props->max_reg_page_list_len = I40IW_MAX_PAGES_PER_FMR; return 0; } diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index 75b2f7d4cd95..10e203a2fd4c 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -537,7 +537,7 @@ static int mlx4_ib_query_device(struct ib_device *ibdev, props->max_srq = dev->dev->quotas.srq; props->max_srq_wr = dev->dev->caps.max_srq_wqes - 1; props->max_srq_sge = dev->dev->caps.max_srq_sge; - props->max_fast_reg_page_list_len = MLX4_MAX_FAST_REG_PAGES; + props->max_reg_page_list_len = MLX4_MAX_FAST_REG_PAGES; props->local_ca_ack_delay = dev->dev->caps.local_ca_ack_delay; props->atomic_cap = dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_ATOMIC ? IB_ATOMIC_HCA : IB_ATOMIC_NONE; diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index 2a55372b7da2..0452c9f9d432 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -728,7 +728,7 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, props->local_ca_ack_delay = MLX5_CAP_GEN(mdev, local_ca_ack_delay); props->max_res_rd_atom = props->max_qp_rd_atom * props->max_qp; props->max_srq_sge = max_rq_sg - 1; - props->max_fast_reg_page_list_len = + props->max_reg_page_list_len = 1 << MLX5_CAP_GEN(mdev, log_max_klm_list_size); get_atomic_caps(dev, props); props->masked_atomic_cap = IB_ATOMIC_NONE; diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c index c197cd9b193f..93a7a72f7c36 100644 --- a/drivers/infiniband/hw/mthca/mthca_provider.c +++ b/drivers/infiniband/hw/mthca/mthca_provider.c @@ -116,6 +116,11 @@ static int mthca_query_device(struct ib_device *ibdev, struct ib_device_attr *pr props->max_mcast_qp_attach = MTHCA_QP_PER_MGM; props->max_total_mcast_qp_attach = props->max_mcast_qp_attach * props->max_mcast_grp; + if (mthca_is_memfree(mdev)) + /* For Arbel, all MTTs must fit in the same page. */ + props->max_reg_page_list_len = PAGE_SIZE / sizeof(u64); + else + props->max_reg_page_list_len = INT_MAX; /* * If Sinai memory key optimization is being used, then only * the 8-bit key portion will change. For other HCAs, the diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c index 2f30bda8457a..d0ac9af99ac7 100644 --- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c +++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c @@ -142,7 +142,7 @@ int ocrdma_query_device(struct ib_device *ibdev, struct ib_device_attr *attr, attr->max_srq_sge = dev->attr.max_srq_sge; attr->max_srq_wr = dev->attr.max_rqe; attr->local_ca_ack_delay = dev->attr.local_ca_ack_delay; - attr->max_fast_reg_page_list_len = dev->attr.max_pages_per_frmr; + attr->max_reg_page_list_len = dev->attr.max_pages_per_frmr; attr->max_pkeys = 1; return 0; } diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c index d6723c365c7f..35691df9cdf2 100644 --- a/drivers/infiniband/hw/qedr/verbs.c +++ b/drivers/infiniband/hw/qedr/verbs.c @@ -163,7 +163,7 @@ int qedr_query_device(struct ib_device *ibdev, attr->max_srq_wr = qattr->max_srq_wr; attr->local_ca_ack_delay = qattr->dev_ack_delay; - attr->max_fast_reg_page_list_len = qattr->max_mr / 8; + attr->max_reg_page_list_len = qattr->max_mr / 8; attr->max_pkeys = QEDR_ROCE_PKEY_MAX; attr->max_ah = qattr->max_ah; diff --git a/drivers/infiniband/hw/qib/qib_verbs.c b/drivers/infiniband/hw/qib/qib_verbs.c index ac42dce7e281..ff92c44083f7 100644 --- a/drivers/infiniband/hw/qib/qib_verbs.c +++ b/drivers/infiniband/hw/qib/qib_verbs.c @@ -1501,6 +1501,7 @@ static void qib_fill_device_attr(struct qib_devdata *dd) rdi->dparms.props.max_total_mcast_qp_attach = rdi->dparms.props.max_mcast_qp_attach * rdi->dparms.props.max_mcast_grp; + rdi->dparms.props.max_reg_page_list_len = UINT_MAX; /* post send table */ dd->verbs_dev.rdi.post_parms = qib_post_parms; } diff --git a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c index 4996984885c2..a173de1761ca 100644 --- a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c +++ b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c @@ -310,7 +310,7 @@ int usnic_ib_query_device(struct ib_device *ibdev, props->max_srq = 0; props->max_srq_wr = 0; props->max_srq_sge = 0; - props->max_fast_reg_page_list_len = 0; + props->max_reg_page_list_len = 0; props->max_mcast_grp = 0; props->max_mcast_qp_attach = 0; props->max_total_mcast_qp_attach = 0; diff --git a/drivers/infiniband/sw/rxe/rxe.c b/drivers/infiniband/sw/rxe/rxe.c index c21c913f911a..e30fd9694902 100644 --- a/drivers/infiniband/sw/rxe/rxe.c +++ b/drivers/infiniband/sw/rxe/rxe.c @@ -124,7 +124,7 @@ static int rxe_init_device_param(struct rxe_dev *rxe) rxe->attr.max_srq = RXE_MAX_SRQ; rxe->attr.max_srq_wr = RXE_MAX_SRQ_WR; rxe->attr.max_srq_sge = RXE_MAX_SRQ_SGE; - rxe->attr.max_fast_reg_page_list_len = RXE_MAX_FMR_PAGE_LIST_LEN; + rxe->attr.max_reg_page_list_len = RXE_MAX_FMR_PAGE_LIST_LEN; rxe->attr.max_pkeys = RXE_MAX_PKEYS; rxe->attr.local_ca_ack_delay = RXE_LOCAL_CA_ACK_DELAY; diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c index c538a38c91ce..28f26ab4585d 100644 --- a/drivers/infiniband/ulp/iser/iser_verbs.c +++ b/drivers/infiniband/ulp/iser/iser_verbs.c @@ -709,7 +709,7 @@ iser_calc_scsi_params(struct iser_conn *iser_conn, sg_tablesize = DIV_ROUND_UP(max_sectors * 512, SIZE_4K); sup_sg_tablesize = min_t(unsigned, ISCSI_ISER_MAX_SG_TABLESIZE, - device->ib_device->attrs.max_fast_reg_page_list_len); + device->ib_device->attrs.max_reg_page_list_len); iser_conn->scsi_sg_tablesize = min(sg_tablesize, sup_sg_tablesize); } diff --git a/drivers/infiniband/ulp/srp/ib_srp.c b/drivers/infiniband/ulp/srp/ib_srp.c index 2354c742caa1..d752100286f9 100644 --- a/drivers/infiniband/ulp/srp/ib_srp.c +++ b/drivers/infiniband/ulp/srp/ib_srp.c @@ -3580,7 +3580,6 @@ static void srp_add_one(struct ib_device *device) struct ib_device_attr *attr = &device->attrs; struct srp_host *host; int mr_page_shift, p; - u64 max_pages_per_mr; unsigned int flags = 0; srp_dev = kzalloc(sizeof(*srp_dev), GFP_KERNEL); @@ -3595,13 +3594,8 @@ static void srp_add_one(struct ib_device *device) mr_page_shift = max(12, ffs(attr->page_size_cap) - 1); srp_dev->mr_page_size = 1 << mr_page_shift; srp_dev->mr_page_mask = ~((u64) srp_dev->mr_page_size - 1); - max_pages_per_mr = attr->max_mr_size; - do_div(max_pages_per_mr, srp_dev->mr_page_size); - pr_debug("%s: %llu / %u = %llu <> %u\n", __func__, - attr->max_mr_size, srp_dev->mr_page_size, - max_pages_per_mr, SRP_MAX_PAGES_PER_MR); srp_dev->max_pages_per_mr = min_t(u64, SRP_MAX_PAGES_PER_MR, - max_pages_per_mr); + attr->max_reg_page_list_len); srp_dev->has_fmr = (device->alloc_fmr && device->dealloc_fmr && device->map_phys_fmr && device->unmap_fmr); @@ -3620,17 +3614,8 @@ static void srp_add_one(struct ib_device *device) (!srp_dev->has_fmr && !srp_dev->has_fr)) flags |= IB_PD_UNSAFE_GLOBAL_RKEY; - if (srp_dev->use_fast_reg) { - srp_dev->max_pages_per_mr = - min_t(u32, srp_dev->max_pages_per_mr, - attr->max_fast_reg_page_list_len); - } - srp_dev->mr_max_size = srp_dev->mr_page_size * - srp_dev->max_pages_per_mr; - pr_debug("%s: mr_page_shift = %d, device->max_mr_size = %#llx, device->max_fast_reg_page_list_len = %u, max_pages_per_mr = %d, mr_max_size = %#x\n", - device->name, mr_page_shift, attr->max_mr_size, - attr->max_fast_reg_page_list_len, - srp_dev->max_pages_per_mr, srp_dev->mr_max_size); + pr_debug("%s: mr_page_shift = %d, max_pages_per_mr = %d\n", + device->name, mr_page_shift, srp_dev->max_pages_per_mr); INIT_LIST_HEAD(&srp_dev->dev_list); diff --git a/drivers/infiniband/ulp/srp/ib_srp.h b/drivers/infiniband/ulp/srp/ib_srp.h index ab9077b81d5a..9f2ada3855a5 100644 --- a/drivers/infiniband/ulp/srp/ib_srp.h +++ b/drivers/infiniband/ulp/srp/ib_srp.h @@ -83,7 +83,6 @@ enum srp_iu_type { /* * @mr_page_mask: HCA memory registration page mask. * @mr_page_size: HCA memory registration page size. - * @mr_max_size: Maximum size in bytes of a single FMR / FR registration * request. */ struct srp_device { @@ -92,7 +91,6 @@ struct srp_device { struct ib_pd *pd; u64 mr_page_mask; int mr_page_size; - int mr_max_size; int max_pages_per_mr; bool has_fmr; bool has_fr; diff --git a/drivers/nvme/host/rdma.c b/drivers/nvme/host/rdma.c index 24397d306d53..22882783bf71 100644 --- a/drivers/nvme/host/rdma.c +++ b/drivers/nvme/host/rdma.c @@ -1586,7 +1586,7 @@ static int nvme_rdma_configure_admin_queue(struct nvme_rdma_ctrl *ctrl) goto out_free_queue; ctrl->max_fr_pages = min_t(u32, NVME_RDMA_MAX_SEGMENTS, - ctrl->device->dev->attrs.max_fast_reg_page_list_len); + ctrl->device->dev->attrs.max_reg_page_list_len); memset(&ctrl->admin_tag_set, 0, sizeof(ctrl->admin_tag_set)); ctrl->admin_tag_set.ops = &nvme_rdma_admin_mq_ops; diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h index 9d4d2a74c95e..91359e5c3961 100644 --- a/include/rdma/ib_verbs.h +++ b/include/rdma/ib_verbs.h @@ -327,7 +327,7 @@ struct ib_device_attr { int max_srq; int max_srq_wr; int max_srq_sge; - unsigned int max_fast_reg_page_list_len; + unsigned int max_reg_page_list_len; u16 max_pkeys; u8 local_ca_ack_delay; int sig_prot_cap; diff --git a/net/sunrpc/xprtrdma/fmr_ops.c b/net/sunrpc/xprtrdma/fmr_ops.c index 59e64025ed96..6000302eede7 100644 --- a/net/sunrpc/xprtrdma/fmr_ops.c +++ b/net/sunrpc/xprtrdma/fmr_ops.c @@ -160,8 +160,12 @@ static int fmr_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep, struct rpcrdma_create_data_internal *cdata) { + struct ib_device_attr *attrs = &ia->ri_device->attrs; + + ia->ri_max_frmr_depth = min_t(unsigned int, RPCRDMA_MAX_DATA_SEGS, + attrs->max_reg_page_list_len); ia->ri_max_segs = max_t(unsigned int, 1, RPCRDMA_MAX_DATA_SEGS / - RPCRDMA_MAX_FMR_SGES); + ia->ri_max_frmr_depth); return 0; } @@ -170,8 +174,10 @@ fmr_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep, static size_t fmr_op_maxpages(struct rpcrdma_xprt *r_xprt) { + struct rpcrdma_ia *ia = &r_xprt->rx_ia; + return min_t(unsigned int, RPCRDMA_MAX_DATA_SEGS, - RPCRDMA_MAX_HDR_SEGS * RPCRDMA_MAX_FMR_SGES); + RPCRDMA_MAX_HDR_SEGS * ia->ri_max_frmr_depth); } /* Use the ib_map_phys_fmr() verb to register a memory region diff --git a/net/sunrpc/xprtrdma/frwr_ops.c b/net/sunrpc/xprtrdma/frwr_ops.c index f81dd93176c0..11effcca0519 100644 --- a/net/sunrpc/xprtrdma/frwr_ops.c +++ b/net/sunrpc/xprtrdma/frwr_ops.c @@ -84,7 +84,7 @@ frwr_is_supported(struct rpcrdma_ia *ia) if (!(attrs->device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS)) goto out_not_supported; - if (attrs->max_fast_reg_page_list_len == 0) + if (attrs->max_reg_page_list_len == 0) goto out_not_supported; return true; @@ -215,7 +215,7 @@ frwr_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep, ia->ri_max_frmr_depth = min_t(unsigned int, RPCRDMA_MAX_DATA_SEGS, - attrs->max_fast_reg_page_list_len); + attrs->max_reg_page_list_len); dprintk("RPC: %s: device's max FR page list len = %u\n", __func__, ia->ri_max_frmr_depth); diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c index a9d9cb1ba4c6..5725abf739ee 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_transport.c +++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c @@ -1001,7 +1001,7 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt) newxprt->sc_reader = rdma_read_chunk_lcl; if (dev->attrs.device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS) { newxprt->sc_frmr_pg_list_len = - dev->attrs.max_fast_reg_page_list_len; + dev->attrs.max_reg_page_list_len; newxprt->sc_dev_caps |= SVCRDMA_DEVCAP_FAST_REG; newxprt->sc_reader = rdma_read_chunk_frmr; } else
Currently FMRs don't even expose maximum supported page list length, which makes it impossible for to know what it actually is. Thus, rename max_fast_reg_page_list_len to max_reg_page_list_len so it is the same capability regardless of registration method is used. Drivers that support both methods should expose the minimum between the two. Signed-off-by: Sagi Grimberg <sagi@grimberg.me> --- drivers/infiniband/core/rw.c | 2 +- drivers/infiniband/hw/bnxt_re/ib_verbs.c | 2 +- drivers/infiniband/hw/cxgb3/iwch_provider.c | 2 +- drivers/infiniband/hw/cxgb4/provider.c | 2 +- drivers/infiniband/hw/hfi1/verbs.c | 2 +- drivers/infiniband/hw/i40iw/i40iw_verbs.c | 2 +- drivers/infiniband/hw/mlx4/main.c | 2 +- drivers/infiniband/hw/mlx5/main.c | 2 +- drivers/infiniband/hw/mthca/mthca_provider.c | 5 +++++ drivers/infiniband/hw/ocrdma/ocrdma_verbs.c | 2 +- drivers/infiniband/hw/qedr/verbs.c | 2 +- drivers/infiniband/hw/qib/qib_verbs.c | 1 + drivers/infiniband/hw/usnic/usnic_ib_verbs.c | 2 +- drivers/infiniband/sw/rxe/rxe.c | 2 +- drivers/infiniband/ulp/iser/iser_verbs.c | 2 +- drivers/infiniband/ulp/srp/ib_srp.c | 21 +++------------------ drivers/infiniband/ulp/srp/ib_srp.h | 2 -- drivers/nvme/host/rdma.c | 2 +- include/rdma/ib_verbs.h | 2 +- net/sunrpc/xprtrdma/fmr_ops.c | 10 ++++++++-- net/sunrpc/xprtrdma/frwr_ops.c | 4 ++-- net/sunrpc/xprtrdma/svc_rdma_transport.c | 2 +- 22 files changed, 35 insertions(+), 40 deletions(-)