From 1dcb6c36a5ebac46099b6363ccf8f4e7563d51e2 Mon Sep 17 00:00:00 2001 From: Eli Cohen Date: Tue, 14 Jul 2020 21:28:32 -0700 Subject: net/mlx5: Support setting access rights of dma addresses mlx5_fill_page_frag_array() is used to populate dma addresses to resources that require it, such as QPs, RQs etc. When the resource is used, PA list permissions are ignored. For resources that use MTT list, the user is required to provide the access rights. Subsequent patches use resources that require MTT lists, so modify API and implementation to support that. Signed-off-by: Eli Cohen Reviewed-by: Parav Pandit Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/alloc.c | 11 +++++++++-- include/linux/mlx5/driver.h | 1 + include/linux/mlx5/mlx5_ifc.h | 6 ++++++ 3 files changed, 16 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/alloc.c b/drivers/net/ethernet/mellanox/mlx5/core/alloc.c index 42198e64a7f4..8db4b5f0f963 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/alloc.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/alloc.c @@ -299,11 +299,18 @@ void mlx5_fill_page_array(struct mlx5_frag_buf *buf, __be64 *pas) } EXPORT_SYMBOL_GPL(mlx5_fill_page_array); -void mlx5_fill_page_frag_array(struct mlx5_frag_buf *buf, __be64 *pas) +void mlx5_fill_page_frag_array_perm(struct mlx5_frag_buf *buf, __be64 *pas, u8 perm) { int i; + WARN_ON(perm & 0xfc); for (i = 0; i < buf->npages; i++) - pas[i] = cpu_to_be64(buf->frags[i].map); + pas[i] = cpu_to_be64(buf->frags[i].map | perm); +} +EXPORT_SYMBOL_GPL(mlx5_fill_page_frag_array_perm); + +void mlx5_fill_page_frag_array(struct mlx5_frag_buf *buf, __be64 *pas) +{ + mlx5_fill_page_frag_array_perm(buf, pas, 0); } EXPORT_SYMBOL_GPL(mlx5_fill_page_frag_array); diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h index 13c0e4556eda..f2557d7e1355 100644 --- a/include/linux/mlx5/driver.h +++ b/include/linux/mlx5/driver.h @@ -971,6 +971,7 @@ void mlx5_register_debugfs(void); void mlx5_unregister_debugfs(void); void mlx5_fill_page_array(struct mlx5_frag_buf *buf, __be64 *pas); +void mlx5_fill_page_frag_array_perm(struct mlx5_frag_buf *buf, __be64 *pas, u8 perm); void mlx5_fill_page_frag_array(struct mlx5_frag_buf *frag_buf, __be64 *pas); int mlx5_vector2eqn(struct mlx5_core_dev *dev, int vector, int *eqn, unsigned int *irqn); diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h index 3786888cb1ba..5890e5c9da77 100644 --- a/include/linux/mlx5/mlx5_ifc.h +++ b/include/linux/mlx5/mlx5_ifc.h @@ -10653,4 +10653,10 @@ struct mlx5_ifc_tls_progress_params_bits { u8 hw_offset_record_number[0x18]; }; +enum { + MLX5_MTT_PERM_READ = 1 << 0, + MLX5_MTT_PERM_WRITE = 1 << 1, + MLX5_MTT_PERM_RW = MLX5_MTT_PERM_READ | MLX5_MTT_PERM_WRITE, +}; + #endif /* MLX5_IFC_H */ -- cgit v1.2.3 From 2a913f23447ce7ef2a4dcaaa230ff43116cf5249 Mon Sep 17 00:00:00 2001 From: Eli Cohen Date: Tue, 14 Jul 2020 21:28:33 -0700 Subject: net/mlx5: Add VDPA interface type to supported enumerations VDPA is a new interface that will be added in subsequent patches. It uses mlx5 core devices and resources. Add an interface type for it. Signed-off-by: Eli Cohen Reviewed-by: Parav Pandit Signed-off-by: Saeed Mahameed --- include/linux/mlx5/driver.h | 1 + 1 file changed, 1 insertion(+) diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h index f2557d7e1355..5ecc48831ae8 100644 --- a/include/linux/mlx5/driver.h +++ b/include/linux/mlx5/driver.h @@ -1054,6 +1054,7 @@ enum { enum { MLX5_INTERFACE_PROTOCOL_IB = 0, MLX5_INTERFACE_PROTOCOL_ETH = 1, + MLX5_INTERFACE_PROTOCOL_VDPA = 2, }; struct mlx5_interface { -- cgit v1.2.3 From 8a06a79b0aa811eee6d56b3cfc738c5d08b0dc74 Mon Sep 17 00:00:00 2001 From: Eli Cohen Date: Tue, 14 Jul 2020 21:28:34 -0700 Subject: net/mlx5: Add interface changes required for VDPA Rename mlx5_ifc_device_virtio_emulation_cap_bits to mlx5_ifc_virtio_emulation_cap_bits to match names produced by the tools producing these auto generated files. In addition missing capabilities that will be required by VDPA implementation. Signed-off-by: Eli Cohen Reviewed-by: Parav Pandit Signed-off-by: Saeed Mahameed --- include/linux/mlx5/device.h | 4 +- include/linux/mlx5/mlx5_ifc.h | 112 ++++++++++++++++++++++++++++++++++++------ 2 files changed, 100 insertions(+), 16 deletions(-) diff --git a/include/linux/mlx5/device.h b/include/linux/mlx5/device.h index 57db125e5802..2aacf9a8ee4d 100644 --- a/include/linux/mlx5/device.h +++ b/include/linux/mlx5/device.h @@ -1361,11 +1361,11 @@ enum mlx5_qcam_feature_groups { MLX5_ADDR_OF(device_event_cap, (mdev)->caps.hca_cur[MLX5_CAP_DEV_EVENT], cap) #define MLX5_CAP_DEV_VDPA_EMULATION(mdev, cap)\ - MLX5_GET(device_virtio_emulation_cap, \ + MLX5_GET(virtio_emulation_cap, \ (mdev)->caps.hca_cur[MLX5_CAP_VDPA_EMULATION], cap) #define MLX5_CAP64_DEV_VDPA_EMULATION(mdev, cap)\ - MLX5_GET64(device_virtio_emulation_cap, \ + MLX5_GET64(virtio_emulation_cap, \ (mdev)->caps.hca_cur[MLX5_CAP_VDPA_EMULATION], cap) #define MLX5_CAP_IPSEC(mdev, cap)\ diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h index 5890e5c9da77..435ab47d5362 100644 --- a/include/linux/mlx5/mlx5_ifc.h +++ b/include/linux/mlx5/mlx5_ifc.h @@ -93,6 +93,7 @@ enum { enum { MLX5_OBJ_TYPE_GENEVE_TLV_OPT = 0x000b, + MLX5_OBJ_TYPE_VIRTIO_NET_Q = 0x000d, MLX5_OBJ_TYPE_MKEY = 0xff01, MLX5_OBJ_TYPE_QP = 0xff02, MLX5_OBJ_TYPE_PSV = 0xff03, @@ -981,17 +982,40 @@ struct mlx5_ifc_device_event_cap_bits { u8 user_unaffiliated_events[4][0x40]; }; -struct mlx5_ifc_device_virtio_emulation_cap_bits { - u8 reserved_at_0[0x20]; +struct mlx5_ifc_virtio_emulation_cap_bits { + u8 desc_tunnel_offload_type[0x1]; + u8 eth_frame_offload_type[0x1]; + u8 virtio_version_1_0[0x1]; + u8 device_features_bits_mask[0xd]; + u8 event_mode[0x8]; + u8 virtio_queue_type[0x8]; - u8 reserved_at_20[0x13]; + u8 max_tunnel_desc[0x10]; + u8 reserved_at_30[0x3]; u8 log_doorbell_stride[0x5]; u8 reserved_at_38[0x3]; u8 log_doorbell_bar_size[0x5]; u8 doorbell_bar_offset[0x40]; - u8 reserved_at_80[0x780]; + u8 max_emulated_devices[0x8]; + u8 max_num_virtio_queues[0x18]; + + u8 reserved_at_a0[0x60]; + + u8 umem_1_buffer_param_a[0x20]; + + u8 umem_1_buffer_param_b[0x20]; + + u8 umem_2_buffer_param_a[0x20]; + + u8 umem_2_buffer_param_b[0x20]; + + u8 umem_3_buffer_param_a[0x20]; + + u8 umem_3_buffer_param_b[0x20]; + + u8 reserved_at_1c0[0x640]; }; enum { @@ -1216,7 +1240,9 @@ struct mlx5_ifc_cmd_hca_cap_bits { u8 max_sgl_for_optimized_performance[0x8]; u8 log_max_cq_sz[0x8]; - u8 reserved_at_d0[0xb]; + u8 reserved_at_d0[0x9]; + u8 virtio_net_device_emualtion_manager[0x1]; + u8 virtio_blk_device_emualtion_manager[0x1]; u8 log_max_cq[0x5]; u8 log_max_eq_sz[0x8]; @@ -2952,7 +2978,7 @@ union mlx5_ifc_hca_cap_union_bits { struct mlx5_ifc_fpga_cap_bits fpga_cap; struct mlx5_ifc_tls_cap_bits tls_cap; struct mlx5_ifc_device_mem_cap_bits device_mem_cap; - struct mlx5_ifc_device_virtio_emulation_cap_bits virtio_emulation_cap; + struct mlx5_ifc_virtio_emulation_cap_bits virtio_emulation_cap; u8 reserved_at_0[0x8000]; }; @@ -3298,15 +3324,18 @@ struct mlx5_ifc_scheduling_context_bits { }; struct mlx5_ifc_rqtc_bits { - u8 reserved_at_0[0xa0]; + u8 reserved_at_0[0xa0]; - u8 reserved_at_a0[0x10]; - u8 rqt_max_size[0x10]; + u8 reserved_at_a0[0x5]; + u8 list_q_type[0x3]; + u8 reserved_at_a8[0x8]; + u8 rqt_max_size[0x10]; - u8 reserved_at_c0[0x10]; - u8 rqt_actual_size[0x10]; + u8 rq_vhca_id_format[0x1]; + u8 reserved_at_c1[0xf]; + u8 rqt_actual_size[0x10]; - u8 reserved_at_e0[0x6a0]; + u8 reserved_at_e0[0x6a0]; struct mlx5_ifc_rq_num_bits rq_num[]; }; @@ -7084,7 +7113,7 @@ struct mlx5_ifc_destroy_mkey_out_bits { struct mlx5_ifc_destroy_mkey_in_bits { u8 opcode[0x10]; - u8 reserved_at_10[0x10]; + u8 uid[0x10]; u8 reserved_at_20[0x10]; u8 op_mod[0x10]; @@ -7782,7 +7811,7 @@ struct mlx5_ifc_create_mkey_out_bits { struct mlx5_ifc_create_mkey_in_bits { u8 opcode[0x10]; - u8 reserved_at_10[0x10]; + u8 uid[0x10]; u8 reserved_at_20[0x10]; u8 op_mod[0x10]; @@ -10312,6 +10341,40 @@ struct mlx5_ifc_create_umem_in_bits { struct mlx5_ifc_umem_bits umem; }; +struct mlx5_ifc_create_umem_out_bits { + u8 status[0x8]; + u8 reserved_at_8[0x18]; + + u8 syndrome[0x20]; + + u8 reserved_at_40[0x8]; + u8 umem_id[0x18]; + + u8 reserved_at_60[0x20]; +}; + +struct mlx5_ifc_destroy_umem_in_bits { + u8 opcode[0x10]; + u8 uid[0x10]; + + u8 reserved_at_20[0x10]; + u8 op_mod[0x10]; + + u8 reserved_at_40[0x8]; + u8 umem_id[0x18]; + + u8 reserved_at_60[0x20]; +}; + +struct mlx5_ifc_destroy_umem_out_bits { + u8 status[0x8]; + u8 reserved_at_8[0x18]; + + u8 syndrome[0x20]; + + u8 reserved_at_40[0x40]; +}; + struct mlx5_ifc_create_uctx_in_bits { u8 opcode[0x10]; u8 reserved_at_10[0x10]; @@ -10324,6 +10387,18 @@ struct mlx5_ifc_create_uctx_in_bits { struct mlx5_ifc_uctx_bits uctx; }; +struct mlx5_ifc_create_uctx_out_bits { + u8 status[0x8]; + u8 reserved_at_8[0x18]; + + u8 syndrome[0x20]; + + u8 reserved_at_40[0x10]; + u8 uid[0x10]; + + u8 reserved_at_60[0x20]; +}; + struct mlx5_ifc_destroy_uctx_in_bits { u8 opcode[0x10]; u8 reserved_at_10[0x10]; @@ -10337,6 +10412,15 @@ struct mlx5_ifc_destroy_uctx_in_bits { u8 reserved_at_60[0x20]; }; +struct mlx5_ifc_destroy_uctx_out_bits { + u8 status[0x8]; + u8 reserved_at_8[0x18]; + + u8 syndrome[0x20]; + + u8 reserved_at_40[0x40]; +}; + struct mlx5_ifc_create_sw_icm_in_bits { struct mlx5_ifc_general_obj_in_cmd_hdr_bits hdr; struct mlx5_ifc_sw_icm_bits sw_icm; -- cgit v1.2.3 From 4c2573e1f6c79525df9094f6ba085fcc35c18c45 Mon Sep 17 00:00:00 2001 From: Michael Guralnik Date: Tue, 14 Jul 2020 21:28:35 -0700 Subject: net/mlx5: Enable count action for rules with allow action Enable the creation of rules with allow and count actions. This enables using counters on egress flow tables. Signed-off-by: Michael Guralnik Reviewed-by: Mark Bloch Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/fs_core.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c index e47a66983935..644fe4c2f0fa 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c @@ -1598,6 +1598,7 @@ static struct mlx5_flow_handle *add_rule_fg(struct mlx5_flow_group *fg, static bool counter_is_valid(u32 action) { return (action & (MLX5_FLOW_CONTEXT_ACTION_DROP | + MLX5_FLOW_CONTEXT_ACTION_ALLOW | MLX5_FLOW_CONTEXT_ACTION_FWD_DEST)); } -- cgit v1.2.3 From 042dd05bddbd84e6a52b337a65d1994003c9b9bb Mon Sep 17 00:00:00 2001 From: Meir Lichtinger Date: Thu, 4 Jun 2020 08:49:38 +0300 Subject: RDMA/mlx5: ConnectX-7 new capabilities to set relaxed ordering by UMR Up to ConnectX-7 setting mkey relaxed ordering read/write attributes by UMR is not supported. ConnectX-7 supports this option, which is indicated by two new HCA capabilities - relaxed_ordering_write_umr and relaxed_ordering_read_umr. Signed-off-by: Meir Lichtinger Signed-off-by: Leon Romanovsky --- include/linux/mlx5/mlx5_ifc.h | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h index 435ab47d5362..0257329431e8 100644 --- a/include/linux/mlx5/mlx5_ifc.h +++ b/include/linux/mlx5/mlx5_ifc.h @@ -1240,7 +1240,9 @@ struct mlx5_ifc_cmd_hca_cap_bits { u8 max_sgl_for_optimized_performance[0x8]; u8 log_max_cq_sz[0x8]; - u8 reserved_at_d0[0x9]; + u8 relaxed_ordering_write_umr[0x1]; + u8 relaxed_ordering_read_umr[0x1]; + u8 reserved_at_d2[0x7]; u8 virtio_net_device_emualtion_manager[0x1]; u8 virtio_blk_device_emualtion_manager[0x1]; u8 log_max_cq[0x5]; -- cgit v1.2.3 From 2224635938814fc63004e30f7c41943812bd6f1c Mon Sep 17 00:00:00 2001 From: Meir Lichtinger Date: Thu, 16 Jul 2020 13:52:47 +0300 Subject: RDMA/mlx5: Use MLX5_SET macro instead of local structure Use generic mlx5 structure defined in mlx5_ifc.h to represent ConnectX device data structures instead of using structure defined specifically for mlx5_ib module. Link: https://lore.kernel.org/r/20200716105248.1423452-3-leon@kernel.org Signed-off-by: Meir Lichtinger Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/hw/mlx5/mlx5_ib.h | 9 --------- drivers/infiniband/hw/mlx5/wr.c | 26 ++++++++++++++++---------- include/linux/mlx5/device.h | 1 - 3 files changed, 16 insertions(+), 20 deletions(-) diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h index 5dbe3eb0d9cb..d9dfe38f4160 100644 --- a/drivers/infiniband/hw/mlx5/mlx5_ib.h +++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h @@ -1423,15 +1423,6 @@ static inline void init_query_mad(struct ib_smp *mad) mad->method = IB_MGMT_METHOD_GET; } -static inline u8 convert_access(int acc) -{ - return (acc & IB_ACCESS_REMOTE_ATOMIC ? MLX5_PERM_ATOMIC : 0) | - (acc & IB_ACCESS_REMOTE_WRITE ? MLX5_PERM_REMOTE_WRITE : 0) | - (acc & IB_ACCESS_REMOTE_READ ? MLX5_PERM_REMOTE_READ : 0) | - (acc & IB_ACCESS_LOCAL_WRITE ? MLX5_PERM_LOCAL_WRITE : 0) | - MLX5_PERM_LOCAL_READ; -} - static inline int is_qp1(enum ib_qp_type qp_type) { return qp_type == MLX5_IB_QPT_HW_GSI; diff --git a/drivers/infiniband/hw/mlx5/wr.c b/drivers/infiniband/hw/mlx5/wr.c index 2c6df1c43b55..e58ecb46f8e3 100644 --- a/drivers/infiniband/hw/mlx5/wr.c +++ b/drivers/infiniband/hw/mlx5/wr.c @@ -383,20 +383,26 @@ static void set_reg_mkey_segment(struct mlx5_mkey_seg *seg, memset(seg, 0, sizeof(*seg)); if (wr->send_flags & MLX5_IB_SEND_UMR_DISABLE_MR) - seg->status = MLX5_MKEY_STATUS_FREE; - - seg->flags = convert_access(umrwr->access_flags); + MLX5_SET(mkc, seg, free, 1); + + MLX5_SET(mkc, seg, a, + !!(umrwr->access_flags & IB_ACCESS_REMOTE_ATOMIC)); + MLX5_SET(mkc, seg, rw, + !!(umrwr->access_flags & IB_ACCESS_REMOTE_WRITE)); + MLX5_SET(mkc, seg, rr, !!(umrwr->access_flags & IB_ACCESS_REMOTE_READ)); + MLX5_SET(mkc, seg, lw, !!(umrwr->access_flags & IB_ACCESS_LOCAL_WRITE)); + MLX5_SET(mkc, seg, lr, 1); if (umrwr->pd) - seg->flags_pd = cpu_to_be32(to_mpd(umrwr->pd)->pdn); + MLX5_SET(mkc, seg, pd, to_mpd(umrwr->pd)->pdn); if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_TRANSLATION && !umrwr->length) - seg->flags_pd |= cpu_to_be32(MLX5_MKEY_LEN64); + MLX5_SET(mkc, seg, length64, 1); - seg->start_addr = cpu_to_be64(umrwr->virt_addr); - seg->len = cpu_to_be64(umrwr->length); - seg->log2_page_size = umrwr->page_shift; - seg->qpn_mkey7_0 = cpu_to_be32(0xffffff00 | - mlx5_mkey_variant(umrwr->mkey)); + MLX5_SET64(mkc, seg, start_addr, umrwr->virt_addr); + MLX5_SET64(mkc, seg, len, umrwr->length); + MLX5_SET(mkc, seg, log_page_size, umrwr->page_shift); + MLX5_SET(mkc, seg, qpn, 0xffffff); + MLX5_SET(mkc, seg, mkey_7_0, mlx5_mkey_variant(umrwr->mkey)); } static void set_reg_data_seg(struct mlx5_wqe_data_seg *dseg, diff --git a/include/linux/mlx5/device.h b/include/linux/mlx5/device.h index 2aacf9a8ee4d..d184b579617f 100644 --- a/include/linux/mlx5/device.h +++ b/include/linux/mlx5/device.h @@ -1007,7 +1007,6 @@ enum { MLX5_MKEY_REMOTE_INVAL = 1 << 24, MLX5_MKEY_FLAG_SYNC_UMR = 1 << 29, MLX5_MKEY_BSF_EN = 1 << 30, - MLX5_MKEY_LEN64 = 1 << 31, }; struct mlx5_mkey_seg { -- cgit v1.2.3 From 896ec9735336f5adb576d372ed7e411bce2fc74c Mon Sep 17 00:00:00 2001 From: Meir Lichtinger Date: Thu, 16 Jul 2020 13:52:48 +0300 Subject: RDMA/mlx5: Set mkey relaxed ordering by UMR with ConnectX-7 Up to ConnectX-7 UMR is not used when user passes relaxed ordering access flag. ConnectX-7 supports setting relaxed ordering read/write mkey attribute by UMR, indicated by new HCA capabilities. With ConnectX-7 driver uses UMR when user set relaxed ordering access flag, in contrast to previous silicon models. Specifically it includes setting relvant flags of mkey context mask in UMR control segment, and relaxed ordering write and read flags in UMR mkey context segment. Link: https://lore.kernel.org/r/20200716105248.1423452-4-leon@kernel.org Signed-off-by: Meir Lichtinger Reviewed-by: Michael Guralnik Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/hw/mlx5/mlx5_ib.h | 9 ++++++-- drivers/infiniband/hw/mlx5/wr.c | 44 ++++++++++++++++++++++++++++-------- include/linux/mlx5/device.h | 4 +++- 3 files changed, 45 insertions(+), 12 deletions(-) diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h index d9dfe38f4160..cdd04b3d7d51 100644 --- a/drivers/infiniband/hw/mlx5/mlx5_ib.h +++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h @@ -1524,8 +1524,13 @@ static inline bool mlx5_ib_can_use_umr(struct mlx5_ib_dev *dev, return false; if (access_flags & IB_ACCESS_RELAXED_ORDERING && - (MLX5_CAP_GEN(dev->mdev, relaxed_ordering_write) || - MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read))) + MLX5_CAP_GEN(dev->mdev, relaxed_ordering_write) && + !MLX5_CAP_GEN(dev->mdev, relaxed_ordering_write_umr)) + return false; + + if (access_flags & IB_ACCESS_RELAXED_ORDERING && + MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read) && + !MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read_umr)) return false; return true; diff --git a/drivers/infiniband/hw/mlx5/wr.c b/drivers/infiniband/hw/mlx5/wr.c index e58ecb46f8e3..4d4f8c22b3e6 100644 --- a/drivers/infiniband/hw/mlx5/wr.c +++ b/drivers/infiniband/hw/mlx5/wr.c @@ -263,7 +263,9 @@ static __be64 get_umr_update_translation_mask(void) return cpu_to_be64(result); } -static __be64 get_umr_update_access_mask(int atomic) +static __be64 get_umr_update_access_mask(int atomic, + int relaxed_ordering_write, + int relaxed_ordering_read) { u64 result; @@ -275,6 +277,12 @@ static __be64 get_umr_update_access_mask(int atomic) if (atomic) result |= MLX5_MKEY_MASK_A; + if (relaxed_ordering_write) + result |= MLX5_MKEY_MASK_RELAXED_ORDERING_WRITE; + + if (relaxed_ordering_read) + result |= MLX5_MKEY_MASK_RELAXED_ORDERING_READ; + return cpu_to_be64(result); } @@ -289,17 +297,28 @@ static __be64 get_umr_update_pd_mask(void) static int umr_check_mkey_mask(struct mlx5_ib_dev *dev, u64 mask) { - if ((mask & MLX5_MKEY_MASK_PAGE_SIZE && - MLX5_CAP_GEN(dev->mdev, umr_modify_entity_size_disabled)) || - (mask & MLX5_MKEY_MASK_A && - MLX5_CAP_GEN(dev->mdev, umr_modify_atomic_disabled))) + if (mask & MLX5_MKEY_MASK_PAGE_SIZE && + MLX5_CAP_GEN(dev->mdev, umr_modify_entity_size_disabled)) + return -EPERM; + + if (mask & MLX5_MKEY_MASK_A && + MLX5_CAP_GEN(dev->mdev, umr_modify_atomic_disabled)) + return -EPERM; + + if (mask & MLX5_MKEY_MASK_RELAXED_ORDERING_WRITE && + !MLX5_CAP_GEN(dev->mdev, relaxed_ordering_write_umr)) + return -EPERM; + + if (mask & MLX5_MKEY_MASK_RELAXED_ORDERING_READ && + !MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read_umr)) return -EPERM; + return 0; } static int set_reg_umr_segment(struct mlx5_ib_dev *dev, struct mlx5_wqe_umr_ctrl_seg *umr, - const struct ib_send_wr *wr, int atomic) + const struct ib_send_wr *wr) { const struct mlx5_umr_wr *umrwr = umr_wr(wr); @@ -325,7 +344,10 @@ static int set_reg_umr_segment(struct mlx5_ib_dev *dev, if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_TRANSLATION) umr->mkey_mask |= get_umr_update_translation_mask(); if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_PD_ACCESS) { - umr->mkey_mask |= get_umr_update_access_mask(atomic); + umr->mkey_mask |= get_umr_update_access_mask( + !!(MLX5_CAP_GEN(dev->mdev, atomic)), + !!(MLX5_CAP_GEN(dev->mdev, relaxed_ordering_write_umr)), + !!(MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read_umr))); umr->mkey_mask |= get_umr_update_pd_mask(); } if (wr->send_flags & MLX5_IB_SEND_UMR_ENABLE_MR) @@ -392,6 +414,11 @@ static void set_reg_mkey_segment(struct mlx5_mkey_seg *seg, MLX5_SET(mkc, seg, rr, !!(umrwr->access_flags & IB_ACCESS_REMOTE_READ)); MLX5_SET(mkc, seg, lw, !!(umrwr->access_flags & IB_ACCESS_LOCAL_WRITE)); MLX5_SET(mkc, seg, lr, 1); + MLX5_SET(mkc, seg, relaxed_ordering_write, + !!(umrwr->access_flags & IB_ACCESS_RELAXED_ORDERING)); + MLX5_SET(mkc, seg, relaxed_ordering_read, + !!(umrwr->access_flags & IB_ACCESS_RELAXED_ORDERING)); + if (umrwr->pd) MLX5_SET(mkc, seg, pd, to_mpd(umrwr->pd)->pdn); if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_TRANSLATION && @@ -1230,8 +1257,7 @@ static int handle_qpt_reg_umr(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, qp->sq.wr_data[idx] = MLX5_IB_WR_UMR; (*ctrl)->imm = cpu_to_be32(umr_wr(wr)->mkey); - err = set_reg_umr_segment(dev, *seg, wr, - !!(MLX5_CAP_GEN(dev->mdev, atomic))); + err = set_reg_umr_segment(dev, *seg, wr); if (unlikely(err)) goto out; *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); diff --git a/include/linux/mlx5/device.h b/include/linux/mlx5/device.h index d184b579617f..4d3376e20f5e 100644 --- a/include/linux/mlx5/device.h +++ b/include/linux/mlx5/device.h @@ -276,7 +276,9 @@ enum { MLX5_MKEY_MASK_RW = 1ull << 20, MLX5_MKEY_MASK_A = 1ull << 21, MLX5_MKEY_MASK_SMALL_FENCE = 1ull << 23, - MLX5_MKEY_MASK_FREE = 1ull << 29, + MLX5_MKEY_MASK_RELAXED_ORDERING_WRITE = 1ull << 25, + MLX5_MKEY_MASK_FREE = 1ull << 29, + MLX5_MKEY_MASK_RELAXED_ORDERING_READ = 1ull << 47, }; enum { -- cgit v1.2.3