From 17793833f81ceb319be599ec09498ec0136d9acf Mon Sep 17 00:00:00 2001 From: Leon Romanovsky Date: Mon, 4 May 2020 16:25:41 +0300 Subject: RDMA/ucma: Return stable IB device index as identifier The librdmacm uses node_guid as identifier to correlate between IB devices and CMA devices. However FW resets cause to such "connection" to be lost and require from the user to restart its application. Extend UCMA to return IB device index, which is stable identifier. Link: https://lore.kernel.org/r/20200504132541.355710-1-leon@kernel.org Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- include/uapi/rdma/rdma_user_cm.h | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'include/uapi') diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index e42940a215a3..1bb6e75d254b 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -164,6 +164,8 @@ struct rdma_ucm_query_route_resp { __u32 num_paths; __u8 port_num; __u8 reserved[3]; + __u32 ibdev_index; + __u32 reserved1; }; struct rdma_ucm_query_addr_resp { @@ -175,6 +177,8 @@ struct rdma_ucm_query_addr_resp { __u16 dst_size; struct __kernel_sockaddr_storage src_addr; struct __kernel_sockaddr_storage dst_addr; + __u32 ibdev_index; + __u32 reserved1; }; struct rdma_ucm_query_path_resp { -- cgit v1.2.3-59-g8ed1b From 8c112a5f29a343f89072bed4b9fa176fea226798 Mon Sep 17 00:00:00 2001 From: Maor Gottlieb Date: Mon, 4 May 2020 08:30:12 +0300 Subject: RDMA/mlx5: Add support in steering default miss User can configure default miss rule in order to skip matching in the user domain and forward the packet to the kernel steering domain. When user requests a default miss rule, we add steering rule to forward the traffic to the next namespace. Link: https://lore.kernel.org/r/20200504053012.270689-5-leon@kernel.org Signed-off-by: Maor Gottlieb Reviewed-by: Mark Zhang Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/hw/mlx5/flow.c | 32 ++++++++++++++++++++++++++------ drivers/infiniband/hw/mlx5/main.c | 9 ++++----- include/uapi/rdma/mlx5_user_ioctl_cmds.h | 5 +++++ 3 files changed, 35 insertions(+), 11 deletions(-) (limited to 'include/uapi') diff --git a/drivers/infiniband/hw/mlx5/flow.c b/drivers/infiniband/hw/mlx5/flow.c index 5533b5083c29..3fa66474afa6 100644 --- a/drivers/infiniband/hw/mlx5/flow.c +++ b/drivers/infiniband/hw/mlx5/flow.c @@ -69,19 +69,32 @@ static const struct uverbs_attr_spec mlx5_ib_flow_type[] = { static int get_dests(struct uverbs_attr_bundle *attrs, struct mlx5_ib_flow_matcher *fs_matcher, int *dest_id, - int *dest_type, struct ib_qp **qp) + int *dest_type, struct ib_qp **qp, bool *def_miss) { bool dest_devx, dest_qp; void *devx_obj; + u32 flags; + int err; dest_devx = uverbs_attr_is_valid(attrs, MLX5_IB_ATTR_CREATE_FLOW_DEST_DEVX); dest_qp = uverbs_attr_is_valid(attrs, MLX5_IB_ATTR_CREATE_FLOW_DEST_QP); - if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_BYPASS && - ((dest_devx && dest_qp) || (!dest_devx && !dest_qp))) - return -EINVAL; + *def_miss = false; + err = uverbs_get_flags32(&flags, attrs, + MLX5_IB_ATTR_CREATE_FLOW_FLAGS, + MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DEFAULT_MISS); + if (err) + return err; + *def_miss = flags & MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DEFAULT_MISS; + + if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_BYPASS) { + if (dest_devx && (dest_qp || *def_miss)) + return -EINVAL; + else if (dest_qp && *def_miss) + return -EINVAL; + } /* Allow only DEVX object as dest when inserting to FDB */ if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB && !dest_devx) @@ -153,6 +166,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)( void *devx_obj, *cmd_in; struct ib_uobject *uobj; struct mlx5_ib_dev *dev; + bool def_miss; if (!capable(CAP_NET_RAW)) return -EPERM; @@ -162,9 +176,12 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)( uobj = uverbs_attr_get_uobject(attrs, MLX5_IB_ATTR_CREATE_FLOW_HANDLE); dev = mlx5_udata_to_mdev(&attrs->driver_udata); - if (get_dests(attrs, fs_matcher, &dest_id, &dest_type, &qp)) + if (get_dests(attrs, fs_matcher, &dest_id, &dest_type, &qp, &def_miss)) return -EINVAL; + if (def_miss) + flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_NS; + len = uverbs_attr_get_uobjs_arr(attrs, MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX, &arr_flow_actions); if (len) { @@ -636,7 +653,10 @@ DECLARE_UVERBS_NAMED_METHOD( UVERBS_ATTR_PTR_IN(MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX_OFFSET, UVERBS_ATTR_MIN_SIZE(sizeof(u32)), UA_OPTIONAL, - UA_ALLOC_AND_COPY)); + UA_ALLOC_AND_COPY), + UVERBS_ATTR_FLAGS_IN(MLX5_IB_ATTR_CREATE_FLOW_FLAGS, + enum mlx5_ib_create_flow_flags, + UA_OPTIONAL)); DECLARE_UVERBS_NAMED_METHOD_DESTROY( MLX5_IB_METHOD_DESTROY_FLOW, diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index 3af57dfe7224..38bf3841741c 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -4200,18 +4200,17 @@ mlx5_ib_raw_fs_rule_add(struct mlx5_ib_dev *dev, if (dest_type == MLX5_FLOW_DESTINATION_TYPE_TIR) { dst[dst_num].type = dest_type; - dst[dst_num].tir_num = dest_id; + dst[dst_num++].tir_num = dest_id; flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST; } else if (dest_type == MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE) { dst[dst_num].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE_NUM; - dst[dst_num].ft_num = dest_id; + dst[dst_num++].ft_num = dest_id; flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST; - } else { - dst[dst_num].type = MLX5_FLOW_DESTINATION_TYPE_PORT; + } else if (dest_type == MLX5_FLOW_DESTINATION_TYPE_PORT) { + dst[dst_num++].type = MLX5_FLOW_DESTINATION_TYPE_PORT; flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_ALLOW; } - dst_num++; if (flow_act->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) { dst[dst_num].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER; diff --git a/include/uapi/rdma/mlx5_user_ioctl_cmds.h b/include/uapi/rdma/mlx5_user_ioctl_cmds.h index 24f3388c3182..07cf54333193 100644 --- a/include/uapi/rdma/mlx5_user_ioctl_cmds.h +++ b/include/uapi/rdma/mlx5_user_ioctl_cmds.h @@ -241,6 +241,10 @@ enum mlx5_ib_flow_type { MLX5_IB_FLOW_TYPE_MC_DEFAULT, }; +enum mlx5_ib_create_flow_flags { + MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DEFAULT_MISS = 1 << 0, +}; + enum mlx5_ib_create_flow_attrs { MLX5_IB_ATTR_CREATE_FLOW_HANDLE = (1U << UVERBS_ID_NS_SHIFT), MLX5_IB_ATTR_CREATE_FLOW_MATCH_VALUE, @@ -251,6 +255,7 @@ enum mlx5_ib_create_flow_attrs { MLX5_IB_ATTR_CREATE_FLOW_TAG, MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX, MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX_OFFSET, + MLX5_IB_ATTR_CREATE_FLOW_FLAGS, }; enum mlx5_ib_destoy_flow_attrs { -- cgit v1.2.3-59-g8ed1b From f29de9eee78253d9ae57cd58a6b21eed021742c8 Mon Sep 17 00:00:00 2001 From: Daria Velikovsky Date: Mon, 4 May 2020 08:42:27 +0300 Subject: RDMA/mlx5: Add support for drop action in DV steering When drop action is used the matching packet will stop processing in steering and will be dropped. This functionality will allow users to drop matching packets. Link: https://lore.kernel.org/r/20200504054227.271486-1-leon@kernel.org Signed-off-by: Daria Velikovsky Reviewed-by: Maor Gottlieb Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/hw/mlx5/flow.c | 35 +++++++++++++++++++------------- include/uapi/rdma/mlx5_user_ioctl_cmds.h | 1 + 2 files changed, 22 insertions(+), 14 deletions(-) (limited to 'include/uapi') diff --git a/drivers/infiniband/hw/mlx5/flow.c b/drivers/infiniband/hw/mlx5/flow.c index 3fa66474afa6..6fa1a510c5d7 100644 --- a/drivers/infiniband/hw/mlx5/flow.c +++ b/drivers/infiniband/hw/mlx5/flow.c @@ -69,11 +69,10 @@ static const struct uverbs_attr_spec mlx5_ib_flow_type[] = { static int get_dests(struct uverbs_attr_bundle *attrs, struct mlx5_ib_flow_matcher *fs_matcher, int *dest_id, - int *dest_type, struct ib_qp **qp, bool *def_miss) + int *dest_type, struct ib_qp **qp, u32 *flags) { bool dest_devx, dest_qp; void *devx_obj; - u32 flags; int err; dest_devx = uverbs_attr_is_valid(attrs, @@ -81,23 +80,28 @@ static int get_dests(struct uverbs_attr_bundle *attrs, dest_qp = uverbs_attr_is_valid(attrs, MLX5_IB_ATTR_CREATE_FLOW_DEST_QP); - *def_miss = false; - err = uverbs_get_flags32(&flags, attrs, - MLX5_IB_ATTR_CREATE_FLOW_FLAGS, - MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DEFAULT_MISS); + *flags = 0; + err = uverbs_get_flags32(flags, attrs, MLX5_IB_ATTR_CREATE_FLOW_FLAGS, + MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DEFAULT_MISS | + MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DROP); if (err) return err; - *def_miss = flags & MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DEFAULT_MISS; + + /* Both flags are not allowed */ + if (*flags & MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DEFAULT_MISS && + *flags & MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DROP) + return -EINVAL; if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_BYPASS) { - if (dest_devx && (dest_qp || *def_miss)) + if (dest_devx && (dest_qp || *flags)) return -EINVAL; - else if (dest_qp && *def_miss) + else if (dest_qp && *flags) return -EINVAL; } - /* Allow only DEVX object as dest when inserting to FDB */ - if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB && !dest_devx) + /* Allow only DEVX object, drop as dest for FDB */ + if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB && !(dest_devx || + (*flags & MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DROP))) return -EINVAL; /* Allow only DEVX object or QP as dest when inserting to RDMA_RX */ @@ -166,7 +170,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)( void *devx_obj, *cmd_in; struct ib_uobject *uobj; struct mlx5_ib_dev *dev; - bool def_miss; + u32 flags; if (!capable(CAP_NET_RAW)) return -EPERM; @@ -176,12 +180,15 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)( uobj = uverbs_attr_get_uobject(attrs, MLX5_IB_ATTR_CREATE_FLOW_HANDLE); dev = mlx5_udata_to_mdev(&attrs->driver_udata); - if (get_dests(attrs, fs_matcher, &dest_id, &dest_type, &qp, &def_miss)) + if (get_dests(attrs, fs_matcher, &dest_id, &dest_type, &qp, &flags)) return -EINVAL; - if (def_miss) + if (flags & MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DEFAULT_MISS) flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_NS; + if (flags & MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DROP) + flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_DROP; + len = uverbs_attr_get_uobjs_arr(attrs, MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX, &arr_flow_actions); if (len) { diff --git a/include/uapi/rdma/mlx5_user_ioctl_cmds.h b/include/uapi/rdma/mlx5_user_ioctl_cmds.h index 07cf54333193..8e316ef896b5 100644 --- a/include/uapi/rdma/mlx5_user_ioctl_cmds.h +++ b/include/uapi/rdma/mlx5_user_ioctl_cmds.h @@ -243,6 +243,7 @@ enum mlx5_ib_flow_type { enum mlx5_ib_create_flow_flags { MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DEFAULT_MISS = 1 << 0, + MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DROP = 1 << 1, }; enum mlx5_ib_create_flow_attrs { -- cgit v1.2.3-59-g8ed1b From fe810b509c5f62b5b3d5681ea6f5d36349ced979 Mon Sep 17 00:00:00 2001 From: Kaike Wan Date: Mon, 11 May 2020 12:05:41 -0400 Subject: IB/hfi1: Add accelerated IP capability bit The accelerated IP capability bit is added to allow users to control which feature is enabled and disabled. Link: https://lore.kernel.org/r/20200511160541.173205.96870.stgit@awfm-01.aw.intel.com Reviewed-by: Dennis Dalessandro Reviewed-by: Mike Marciniszyn Signed-off-by: Kaike Wan Signed-off-by: Dennis Dalessandro Signed-off-by: Jason Gunthorpe --- drivers/infiniband/hw/hfi1/common.h | 5 +++-- include/uapi/rdma/hfi/hfi1_user.h | 3 ++- 2 files changed, 5 insertions(+), 3 deletions(-) (limited to 'include/uapi') diff --git a/drivers/infiniband/hw/hfi1/common.h b/drivers/infiniband/hw/hfi1/common.h index 40a1ff0c8a8e..1f7107e35a43 100644 --- a/drivers/infiniband/hw/hfi1/common.h +++ b/drivers/infiniband/hw/hfi1/common.h @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015 - 2018 Intel Corporation. + * Copyright(c) 2015 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -149,7 +149,8 @@ HFI1_CAP_NO_INTEGRITY | \ HFI1_CAP_PKEY_CHECK | \ HFI1_CAP_TID_RDMA | \ - HFI1_CAP_OPFN) << \ + HFI1_CAP_OPFN | \ + HFI1_CAP_AIP) << \ HFI1_CAP_USER_SHIFT) /* * Set of capabilities that need to be enabled for kernel context in diff --git a/include/uapi/rdma/hfi/hfi1_user.h b/include/uapi/rdma/hfi/hfi1_user.h index 01ac5853d9ac..d95ef9a2b032 100644 --- a/include/uapi/rdma/hfi/hfi1_user.h +++ b/include/uapi/rdma/hfi/hfi1_user.h @@ -6,7 +6,7 @@ * * GPL LICENSE SUMMARY * - * Copyright(c) 2015 - 2018 Intel Corporation. + * Copyright(c) 2015 - 2020 Intel Corporation. * * This program is free software; you can redistribute it and/or modify * it under the terms of version 2 of the GNU General Public License as @@ -109,6 +109,7 @@ #define HFI1_CAP_OPFN (1UL << 16) /* Enable the OPFN protocol */ #define HFI1_CAP_SDMA_HEAD_CHECK (1UL << 17) /* SDMA head checking */ #define HFI1_CAP_EARLY_CREDIT_RETURN (1UL << 18) /* early credit return */ +#define HFI1_CAP_AIP (1UL << 19) /* Enable accelerated IP */ #define HFI1_RCVHDR_ENTSIZE_2 (1UL << 0) #define HFI1_RCVHDR_ENTSIZE_16 (1UL << 1) -- cgit v1.2.3-59-g8ed1b From cda9ee494248b890973f5d31cf7851c0d21755b9 Mon Sep 17 00:00:00 2001 From: Yishai Hadas Date: Tue, 19 May 2020 10:27:07 +0300 Subject: IB/uverbs: Extend CQ to get its own asynchronous event FD Extend CQ to get its own asynchronous event FD. The event FD is an optional attribute, in case wasn't given the ufile event FD will be used. Link: https://lore.kernel.org/r/20200519072711.257271-4-leon@kernel.org Signed-off-by: Yishai Hadas Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/core/uverbs.h | 18 ++++++++++++++++++ drivers/infiniband/core/uverbs_std_types_cq.c | 9 ++++++--- include/uapi/rdma/ib_user_ioctl_cmds.h | 1 + 3 files changed, 25 insertions(+), 3 deletions(-) (limited to 'include/uapi') diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h index 34c155f88317..53a10479958b 100644 --- a/drivers/infiniband/core/uverbs.h +++ b/drivers/infiniband/core/uverbs.h @@ -297,6 +297,24 @@ static inline u32 make_port_cap_flags(const struct ib_port_attr *attr) return res; } +static inline struct ib_uverbs_async_event_file * +ib_uverbs_get_async_event(struct uverbs_attr_bundle *attrs, + u16 id) +{ + struct ib_uobject *async_ev_file_uobj; + struct ib_uverbs_async_event_file *async_ev_file; + + async_ev_file_uobj = uverbs_attr_get_uobject(attrs, id); + if (IS_ERR(async_ev_file_uobj)) + async_ev_file = READ_ONCE(attrs->ufile->default_async_file); + else + async_ev_file = container_of(async_ev_file_uobj, + struct ib_uverbs_async_event_file, + uobj); + if (async_ev_file) + uverbs_uobject_get(&async_ev_file->uobj); + return async_ev_file; +} void copy_port_attr_to_resp(struct ib_port_attr *attr, struct ib_uverbs_query_port_resp *resp, diff --git a/drivers/infiniband/core/uverbs_std_types_cq.c b/drivers/infiniband/core/uverbs_std_types_cq.c index be534b0af4f8..5dce2c7cc323 100644 --- a/drivers/infiniband/core/uverbs_std_types_cq.c +++ b/drivers/infiniband/core/uverbs_std_types_cq.c @@ -100,9 +100,8 @@ static int UVERBS_HANDLER(UVERBS_METHOD_CQ_CREATE)( uverbs_uobject_get(ev_file_uobj); } - obj->uevent.event_file = READ_ONCE(attrs->ufile->default_async_file); - if (obj->uevent.event_file) - uverbs_uobject_get(&obj->uevent.event_file->uobj); + obj->uevent.event_file = ib_uverbs_get_async_event( + attrs, UVERBS_ATTR_CREATE_CQ_EVENT_FD); if (attr.comp_vector >= attrs->ufile->device->num_comp_vectors) { ret = -EINVAL; @@ -173,6 +172,10 @@ DECLARE_UVERBS_NAMED_METHOD( UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_CQ_RESP_CQE, UVERBS_ATTR_TYPE(u32), UA_MANDATORY), + UVERBS_ATTR_FD(UVERBS_ATTR_CREATE_CQ_EVENT_FD, + UVERBS_OBJECT_ASYNC_EVENT, + UVERBS_ACCESS_READ, + UA_OPTIONAL), UVERBS_ATTR_UHW()); static int UVERBS_HANDLER(UVERBS_METHOD_CQ_DESTROY)( diff --git a/include/uapi/rdma/ib_user_ioctl_cmds.h b/include/uapi/rdma/ib_user_ioctl_cmds.h index d4ddbe4e696c..286fdc1929e0 100644 --- a/include/uapi/rdma/ib_user_ioctl_cmds.h +++ b/include/uapi/rdma/ib_user_ioctl_cmds.h @@ -95,6 +95,7 @@ enum uverbs_attrs_create_cq_cmd_attr_ids { UVERBS_ATTR_CREATE_CQ_COMP_VECTOR, UVERBS_ATTR_CREATE_CQ_FLAGS, UVERBS_ATTR_CREATE_CQ_RESP_CQE, + UVERBS_ATTR_CREATE_CQ_EVENT_FD, }; enum uverbs_attrs_destroy_cq_cmd_attr_ids { -- cgit v1.2.3-59-g8ed1b From 175ba58d62c84e1216cdf8b4f49f79e55e1ed04b Mon Sep 17 00:00:00 2001 From: Yishai Hadas Date: Tue, 19 May 2020 10:27:08 +0300 Subject: IB/uverbs: Move QP, SRQ, WQ type and flags to UAPI These constants are going to be used in the ioctl interface in coming patches so they are part of the UAPI, place them in the correct header for clarity. Link: https://lore.kernel.org/r/20200519072711.257271-5-leon@kernel.org Signed-off-by: Yishai Hadas Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- include/rdma/ib_verbs.h | 43 ++++++++++++++++++--------------- include/uapi/rdma/ib_user_ioctl_verbs.h | 34 ++++++++++++++++++++++++++ 2 files changed, 58 insertions(+), 19 deletions(-) (limited to 'include/uapi') diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h index c988e9205cf9..94533ae16697 100644 --- a/include/rdma/ib_verbs.h +++ b/include/rdma/ib_verbs.h @@ -1040,9 +1040,9 @@ enum ib_cq_notify_flags { }; enum ib_srq_type { - IB_SRQT_BASIC, - IB_SRQT_XRC, - IB_SRQT_TM, + IB_SRQT_BASIC = IB_UVERBS_SRQT_BASIC, + IB_SRQT_XRC = IB_UVERBS_SRQT_XRC, + IB_SRQT_TM = IB_UVERBS_SRQT_TM, }; static inline bool ib_srq_has_cq(enum ib_srq_type srq_type) @@ -1111,16 +1111,16 @@ enum ib_qp_type { IB_QPT_SMI, IB_QPT_GSI, - IB_QPT_RC, - IB_QPT_UC, - IB_QPT_UD, + IB_QPT_RC = IB_UVERBS_QPT_RC, + IB_QPT_UC = IB_UVERBS_QPT_UC, + IB_QPT_UD = IB_UVERBS_QPT_UD, IB_QPT_RAW_IPV6, IB_QPT_RAW_ETHERTYPE, - IB_QPT_RAW_PACKET = 8, - IB_QPT_XRC_INI = 9, - IB_QPT_XRC_TGT, + IB_QPT_RAW_PACKET = IB_UVERBS_QPT_RAW_PACKET, + IB_QPT_XRC_INI = IB_UVERBS_QPT_XRC_INI, + IB_QPT_XRC_TGT = IB_UVERBS_QPT_XRC_TGT, IB_QPT_MAX, - IB_QPT_DRIVER = 0xFF, + IB_QPT_DRIVER = IB_UVERBS_QPT_DRIVER, /* Reserve a range for qp types internal to the low level driver. * These qp types will not be visible at the IB core layer, so the * IB_QPT_MAX usages should not be affected in the core layer @@ -1139,17 +1139,21 @@ enum ib_qp_type { enum ib_qp_create_flags { IB_QP_CREATE_IPOIB_UD_LSO = 1 << 0, - IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK = 1 << 1, + IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK = + IB_UVERBS_QP_CREATE_BLOCK_MULTICAST_LOOPBACK, IB_QP_CREATE_CROSS_CHANNEL = 1 << 2, IB_QP_CREATE_MANAGED_SEND = 1 << 3, IB_QP_CREATE_MANAGED_RECV = 1 << 4, IB_QP_CREATE_NETIF_QP = 1 << 5, IB_QP_CREATE_INTEGRITY_EN = 1 << 6, IB_QP_CREATE_NETDEV_USE = 1 << 7, - IB_QP_CREATE_SCATTER_FCS = 1 << 8, - IB_QP_CREATE_CVLAN_STRIPPING = 1 << 9, + IB_QP_CREATE_SCATTER_FCS = + IB_UVERBS_QP_CREATE_SCATTER_FCS, + IB_QP_CREATE_CVLAN_STRIPPING = + IB_UVERBS_QP_CREATE_CVLAN_STRIPPING, IB_QP_CREATE_SOURCE_QPN = 1 << 10, - IB_QP_CREATE_PCI_WRITE_END_PADDING = 1 << 11, + IB_QP_CREATE_PCI_WRITE_END_PADDING = + IB_UVERBS_QP_CREATE_PCI_WRITE_END_PADDING, /* reserve bits 26-31 for low level drivers' internal use */ IB_QP_CREATE_RESERVED_START = 1 << 26, IB_QP_CREATE_RESERVED_END = 1 << 31, @@ -1654,7 +1658,7 @@ enum ib_raw_packet_caps { }; enum ib_wq_type { - IB_WQT_RQ + IB_WQT_RQ = IB_UVERBS_WQT_RQ, }; enum ib_wq_state { @@ -1677,10 +1681,11 @@ struct ib_wq { }; enum ib_wq_flags { - IB_WQ_FLAGS_CVLAN_STRIPPING = 1 << 0, - IB_WQ_FLAGS_SCATTER_FCS = 1 << 1, - IB_WQ_FLAGS_DELAY_DROP = 1 << 2, - IB_WQ_FLAGS_PCI_WRITE_END_PADDING = 1 << 3, + IB_WQ_FLAGS_CVLAN_STRIPPING = IB_UVERBS_WQ_FLAGS_CVLAN_STRIPPING, + IB_WQ_FLAGS_SCATTER_FCS = IB_UVERBS_WQ_FLAGS_SCATTER_FCS, + IB_WQ_FLAGS_DELAY_DROP = IB_UVERBS_WQ_FLAGS_DELAY_DROP, + IB_WQ_FLAGS_PCI_WRITE_END_PADDING = + IB_UVERBS_WQ_FLAGS_PCI_WRITE_END_PADDING, }; struct ib_wq_init_attr { diff --git a/include/uapi/rdma/ib_user_ioctl_verbs.h b/include/uapi/rdma/ib_user_ioctl_verbs.h index a640bb814be0..b1662dfe86a6 100644 --- a/include/uapi/rdma/ib_user_ioctl_verbs.h +++ b/include/uapi/rdma/ib_user_ioctl_verbs.h @@ -64,6 +64,40 @@ enum ib_uverbs_access_flags { ~(IB_UVERBS_ACCESS_OPTIONAL_FIRST - 1) }; +enum ib_uverbs_srq_type { + IB_UVERBS_SRQT_BASIC, + IB_UVERBS_SRQT_XRC, + IB_UVERBS_SRQT_TM, +}; + +enum ib_uverbs_wq_type { + IB_UVERBS_WQT_RQ, +}; + +enum ib_uverbs_wq_flags { + IB_UVERBS_WQ_FLAGS_CVLAN_STRIPPING = 1 << 0, + IB_UVERBS_WQ_FLAGS_SCATTER_FCS = 1 << 1, + IB_UVERBS_WQ_FLAGS_DELAY_DROP = 1 << 2, + IB_UVERBS_WQ_FLAGS_PCI_WRITE_END_PADDING = 1 << 3, +}; + +enum ib_uverbs_qp_type { + IB_UVERBS_QPT_RC = 2, + IB_UVERBS_QPT_UC, + IB_UVERBS_QPT_UD, + IB_UVERBS_QPT_RAW_PACKET = 8, + IB_UVERBS_QPT_XRC_INI, + IB_UVERBS_QPT_XRC_TGT, + IB_UVERBS_QPT_DRIVER = 0xFF, +}; + +enum ib_uverbs_qp_create_flags { + IB_UVERBS_QP_CREATE_BLOCK_MULTICAST_LOOPBACK = 1 << 1, + IB_UVERBS_QP_CREATE_SCATTER_FCS = 1 << 8, + IB_UVERBS_QP_CREATE_CVLAN_STRIPPING = 1 << 9, + IB_UVERBS_QP_CREATE_PCI_WRITE_END_PADDING = 1 << 11, +}; + enum ib_uverbs_query_port_cap_flags { IB_UVERBS_PCF_SM = 1 << 1, IB_UVERBS_PCF_NOTICE_SUP = 1 << 2, -- cgit v1.2.3-59-g8ed1b From c3eab946aba443f0b44a08f446735c74495610a9 Mon Sep 17 00:00:00 2001 From: Yishai Hadas Date: Tue, 19 May 2020 10:27:09 +0300 Subject: IB/uverbs: Introduce create/destroy SRQ commands over ioctl Introduce create/destroy SRQ commands over the ioctl interface to let it be extended to get an asynchronous event FD. Link: https://lore.kernel.org/r/20200519072711.257271-6-leon@kernel.org Signed-off-by: Yishai Hadas Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/core/Makefile | 3 +- drivers/infiniband/core/rdma_core.h | 1 + drivers/infiniband/core/uverbs_std_types.c | 32 ---- drivers/infiniband/core/uverbs_std_types_srq.c | 234 +++++++++++++++++++++++++ drivers/infiniband/core/uverbs_uapi.c | 1 + include/uapi/rdma/ib_user_ioctl_cmds.h | 27 +++ 6 files changed, 265 insertions(+), 33 deletions(-) create mode 100644 drivers/infiniband/core/uverbs_std_types_srq.c (limited to 'include/uapi') diff --git a/drivers/infiniband/core/Makefile b/drivers/infiniband/core/Makefile index 870f0fcd54d5..d7b46a7c07fd 100644 --- a/drivers/infiniband/core/Makefile +++ b/drivers/infiniband/core/Makefile @@ -36,6 +36,7 @@ ib_uverbs-y := uverbs_main.o uverbs_cmd.o uverbs_marshall.o \ uverbs_std_types_flow_action.o uverbs_std_types_dm.o \ uverbs_std_types_mr.o uverbs_std_types_counters.o \ uverbs_uapi.o uverbs_std_types_device.o \ - uverbs_std_types_async_fd.o + uverbs_std_types_async_fd.o \ + uverbs_std_types_srq.o ib_uverbs-$(CONFIG_INFINIBAND_USER_MEM) += umem.o ib_uverbs-$(CONFIG_INFINIBAND_ON_DEMAND_PAGING) += umem_odp.o diff --git a/drivers/infiniband/core/rdma_core.h b/drivers/infiniband/core/rdma_core.h index 2b529233e159..d623f911b70b 100644 --- a/drivers/infiniband/core/rdma_core.h +++ b/drivers/infiniband/core/rdma_core.h @@ -159,6 +159,7 @@ extern const struct uapi_definition uverbs_def_obj_dm[]; extern const struct uapi_definition uverbs_def_obj_flow_action[]; extern const struct uapi_definition uverbs_def_obj_intf[]; extern const struct uapi_definition uverbs_def_obj_mr[]; +extern const struct uapi_definition uverbs_def_obj_srq[]; extern const struct uapi_definition uverbs_def_write_intf[]; static inline const struct uverbs_api_write_method * diff --git a/drivers/infiniband/core/uverbs_std_types.c b/drivers/infiniband/core/uverbs_std_types.c index 3abfc63225cb..d9b6912eafa8 100644 --- a/drivers/infiniband/core/uverbs_std_types.c +++ b/drivers/infiniband/core/uverbs_std_types.c @@ -142,31 +142,6 @@ static int uverbs_free_wq(struct ib_uobject *uobject, return ret; } -static int uverbs_free_srq(struct ib_uobject *uobject, - enum rdma_remove_reason why, - struct uverbs_attr_bundle *attrs) -{ - struct ib_srq *srq = uobject->object; - struct ib_uevent_object *uevent = - container_of(uobject, struct ib_uevent_object, uobject); - enum ib_srq_type srq_type = srq->srq_type; - int ret; - - ret = ib_destroy_srq_user(srq, &attrs->driver_udata); - if (ib_is_destroy_retryable(ret, why, uobject)) - return ret; - - if (srq_type == IB_SRQT_XRC) { - struct ib_usrq_object *us = - container_of(uevent, struct ib_usrq_object, uevent); - - atomic_dec(&us->uxrcd->refcnt); - } - - ib_uverbs_release_uevent(uevent); - return ret; -} - static int uverbs_free_xrcd(struct ib_uobject *uobject, enum rdma_remove_reason why, struct uverbs_attr_bundle *attrs) @@ -267,11 +242,6 @@ DECLARE_UVERBS_NAMED_OBJECT(UVERBS_OBJECT_MW, UVERBS_TYPE_ALLOC_IDR(uverbs_free_mw), &UVERBS_METHOD(UVERBS_METHOD_MW_DESTROY)); -DECLARE_UVERBS_NAMED_OBJECT( - UVERBS_OBJECT_SRQ, - UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_usrq_object), - uverbs_free_srq)); - DECLARE_UVERBS_NAMED_METHOD_DESTROY( UVERBS_METHOD_AH_DESTROY, UVERBS_ATTR_IDR(UVERBS_ATTR_DESTROY_AH_HANDLE, @@ -346,8 +316,6 @@ const struct uapi_definition uverbs_def_obj_intf[] = { UAPI_DEF_OBJ_NEEDS_FN(destroy_ah)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_MW, UAPI_DEF_OBJ_NEEDS_FN(dealloc_mw)), - UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_SRQ, - UAPI_DEF_OBJ_NEEDS_FN(destroy_srq)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_FLOW, UAPI_DEF_OBJ_NEEDS_FN(destroy_flow)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_WQ, diff --git a/drivers/infiniband/core/uverbs_std_types_srq.c b/drivers/infiniband/core/uverbs_std_types_srq.c new file mode 100644 index 000000000000..c0ecbba26bf4 --- /dev/null +++ b/drivers/infiniband/core/uverbs_std_types_srq.c @@ -0,0 +1,234 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. + */ + +#include +#include "rdma_core.h" +#include "uverbs.h" + +static int uverbs_free_srq(struct ib_uobject *uobject, + enum rdma_remove_reason why, + struct uverbs_attr_bundle *attrs) +{ + struct ib_srq *srq = uobject->object; + struct ib_uevent_object *uevent = + container_of(uobject, struct ib_uevent_object, uobject); + enum ib_srq_type srq_type = srq->srq_type; + int ret; + + ret = ib_destroy_srq_user(srq, &attrs->driver_udata); + if (ib_is_destroy_retryable(ret, why, uobject)) + return ret; + + if (srq_type == IB_SRQT_XRC) { + struct ib_usrq_object *us = + container_of(uobject, struct ib_usrq_object, + uevent.uobject); + + atomic_dec(&us->uxrcd->refcnt); + } + + ib_uverbs_release_uevent(uevent); + return ret; +} + +static int UVERBS_HANDLER(UVERBS_METHOD_SRQ_CREATE)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_usrq_object *obj = container_of( + uverbs_attr_get_uobject(attrs, UVERBS_ATTR_CREATE_SRQ_HANDLE), + typeof(*obj), uevent.uobject); + struct ib_pd *pd = + uverbs_attr_get_obj(attrs, UVERBS_ATTR_CREATE_SRQ_PD_HANDLE); + struct ib_srq_init_attr attr = {}; + struct ib_uobject *xrcd_uobj; + struct ib_srq *srq; + u64 user_handle; + int ret; + + ret = uverbs_copy_from(&attr.attr.max_sge, attrs, + UVERBS_ATTR_CREATE_SRQ_MAX_SGE); + if (!ret) + ret = uverbs_copy_from(&attr.attr.max_wr, attrs, + UVERBS_ATTR_CREATE_SRQ_MAX_WR); + if (!ret) + ret = uverbs_copy_from(&attr.attr.srq_limit, attrs, + UVERBS_ATTR_CREATE_SRQ_LIMIT); + if (!ret) + ret = uverbs_copy_from(&user_handle, attrs, + UVERBS_ATTR_CREATE_SRQ_USER_HANDLE); + if (!ret) + ret = uverbs_get_const(&attr.srq_type, attrs, + UVERBS_ATTR_CREATE_SRQ_TYPE); + if (ret) + return ret; + + if (ib_srq_has_cq(attr.srq_type)) { + attr.ext.cq = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_SRQ_CQ_HANDLE); + if (IS_ERR(attr.ext.cq)) + return PTR_ERR(attr.ext.cq); + } + + switch (attr.srq_type) { + case IB_UVERBS_SRQT_XRC: + xrcd_uobj = uverbs_attr_get_uobject(attrs, + UVERBS_ATTR_CREATE_SRQ_XRCD_HANDLE); + if (IS_ERR(xrcd_uobj)) + return PTR_ERR(xrcd_uobj); + + attr.ext.xrc.xrcd = (struct ib_xrcd *)xrcd_uobj->object; + if (!attr.ext.xrc.xrcd) + return -EINVAL; + obj->uxrcd = container_of(xrcd_uobj, struct ib_uxrcd_object, + uobject); + atomic_inc(&obj->uxrcd->refcnt); + break; + case IB_UVERBS_SRQT_TM: + ret = uverbs_copy_from(&attr.ext.tag_matching.max_num_tags, + attrs, + UVERBS_ATTR_CREATE_SRQ_MAX_NUM_TAGS); + if (ret) + return ret; + break; + case IB_UVERBS_SRQT_BASIC: + break; + default: + return -EINVAL; + } + + obj->uevent.event_file = ib_uverbs_get_async_event(attrs, + UVERBS_ATTR_CREATE_SRQ_EVENT_FD); + INIT_LIST_HEAD(&obj->uevent.event_list); + attr.event_handler = ib_uverbs_srq_event_handler; + obj->uevent.uobject.user_handle = user_handle; + + srq = ib_create_srq_user(pd, &attr, obj, &attrs->driver_udata); + if (IS_ERR(srq)) { + ret = PTR_ERR(srq); + goto err; + } + + obj->uevent.uobject.object = srq; + uverbs_finalize_uobj_create(attrs, UVERBS_ATTR_CREATE_SRQ_HANDLE); + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_SRQ_RESP_MAX_WR, + &attr.attr.max_wr, + sizeof(attr.attr.max_wr)); + if (ret) + return ret; + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_SRQ_RESP_MAX_SGE, + &attr.attr.max_sge, + sizeof(attr.attr.max_sge)); + if (ret) + return ret; + + if (attr.srq_type == IB_SRQT_XRC) { + ret = uverbs_copy_to(attrs, + UVERBS_ATTR_CREATE_SRQ_RESP_SRQ_NUM, + &srq->ext.xrc.srq_num, + sizeof(srq->ext.xrc.srq_num)); + if (ret) + return ret; + } + + return 0; +err: + if (obj->uevent.event_file) + uverbs_uobject_put(&obj->uevent.event_file->uobj); + if (attr.srq_type == IB_SRQT_XRC) + atomic_dec(&obj->uxrcd->refcnt); + return ret; +}; + +DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_SRQ_CREATE, + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_SRQ_HANDLE, + UVERBS_OBJECT_SRQ, + UVERBS_ACCESS_NEW, + UA_MANDATORY), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_SRQ_PD_HANDLE, + UVERBS_OBJECT_PD, + UVERBS_ACCESS_READ, + UA_MANDATORY), + UVERBS_ATTR_CONST_IN(UVERBS_ATTR_CREATE_SRQ_TYPE, + enum ib_uverbs_srq_type, + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_SRQ_USER_HANDLE, + UVERBS_ATTR_TYPE(u64), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_SRQ_MAX_WR, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_SRQ_MAX_SGE, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_SRQ_LIMIT, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_SRQ_XRCD_HANDLE, + UVERBS_OBJECT_XRCD, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_SRQ_CQ_HANDLE, + UVERBS_OBJECT_CQ, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_SRQ_MAX_NUM_TAGS, + UVERBS_ATTR_TYPE(u32), + UA_OPTIONAL), + UVERBS_ATTR_FD(UVERBS_ATTR_CREATE_SRQ_EVENT_FD, + UVERBS_OBJECT_ASYNC_EVENT, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_SRQ_RESP_MAX_WR, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_SRQ_RESP_MAX_SGE, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_SRQ_RESP_SRQ_NUM, + UVERBS_ATTR_TYPE(u32), + UA_OPTIONAL), + UVERBS_ATTR_UHW()); + +static int UVERBS_HANDLER(UVERBS_METHOD_SRQ_DESTROY)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uobject *uobj = + uverbs_attr_get_uobject(attrs, UVERBS_ATTR_DESTROY_SRQ_HANDLE); + struct ib_usrq_object *obj = + container_of(uobj, struct ib_usrq_object, uevent.uobject); + struct ib_uverbs_destroy_srq_resp resp = { + .events_reported = obj->uevent.events_reported + }; + + return uverbs_copy_to(attrs, UVERBS_ATTR_DESTROY_SRQ_RESP, &resp, + sizeof(resp)); +} + +DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_SRQ_DESTROY, + UVERBS_ATTR_IDR(UVERBS_ATTR_DESTROY_SRQ_HANDLE, + UVERBS_OBJECT_SRQ, + UVERBS_ACCESS_DESTROY, + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_DESTROY_SRQ_RESP, + UVERBS_ATTR_TYPE(struct ib_uverbs_destroy_srq_resp), + UA_MANDATORY)); + +DECLARE_UVERBS_NAMED_OBJECT( + UVERBS_OBJECT_SRQ, + UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_usrq_object), + uverbs_free_srq), + &UVERBS_METHOD(UVERBS_METHOD_SRQ_CREATE), + &UVERBS_METHOD(UVERBS_METHOD_SRQ_DESTROY) +); + +const struct uapi_definition uverbs_def_obj_srq[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_SRQ, + UAPI_DEF_OBJ_NEEDS_FN(destroy_srq)), + {} +}; diff --git a/drivers/infiniband/core/uverbs_uapi.c b/drivers/infiniband/core/uverbs_uapi.c index 3f121ac31e0a..3f5627954fe7 100644 --- a/drivers/infiniband/core/uverbs_uapi.c +++ b/drivers/infiniband/core/uverbs_uapi.c @@ -634,6 +634,7 @@ static const struct uapi_definition uverbs_core_api[] = { UAPI_DEF_CHAIN(uverbs_def_obj_flow_action), UAPI_DEF_CHAIN(uverbs_def_obj_intf), UAPI_DEF_CHAIN(uverbs_def_obj_mr), + UAPI_DEF_CHAIN(uverbs_def_obj_srq), UAPI_DEF_CHAIN(uverbs_def_write_intf), {}, }; diff --git a/include/uapi/rdma/ib_user_ioctl_cmds.h b/include/uapi/rdma/ib_user_ioctl_cmds.h index 286fdc1929e0..c07af46ff04c 100644 --- a/include/uapi/rdma/ib_user_ioctl_cmds.h +++ b/include/uapi/rdma/ib_user_ioctl_cmds.h @@ -121,6 +121,33 @@ enum uverbs_attrs_destroy_flow_action_esp { UVERBS_ATTR_DESTROY_FLOW_ACTION_HANDLE, }; +enum uverbs_attrs_create_srq_cmd_attr_ids { + UVERBS_ATTR_CREATE_SRQ_HANDLE, + UVERBS_ATTR_CREATE_SRQ_PD_HANDLE, + UVERBS_ATTR_CREATE_SRQ_XRCD_HANDLE, + UVERBS_ATTR_CREATE_SRQ_CQ_HANDLE, + UVERBS_ATTR_CREATE_SRQ_USER_HANDLE, + UVERBS_ATTR_CREATE_SRQ_MAX_WR, + UVERBS_ATTR_CREATE_SRQ_MAX_SGE, + UVERBS_ATTR_CREATE_SRQ_LIMIT, + UVERBS_ATTR_CREATE_SRQ_MAX_NUM_TAGS, + UVERBS_ATTR_CREATE_SRQ_TYPE, + UVERBS_ATTR_CREATE_SRQ_EVENT_FD, + UVERBS_ATTR_CREATE_SRQ_RESP_MAX_WR, + UVERBS_ATTR_CREATE_SRQ_RESP_MAX_SGE, + UVERBS_ATTR_CREATE_SRQ_RESP_SRQ_NUM, +}; + +enum uverbs_attrs_destroy_srq_cmd_attr_ids { + UVERBS_ATTR_DESTROY_SRQ_HANDLE, + UVERBS_ATTR_DESTROY_SRQ_RESP, +}; + +enum uverbs_methods_srq { + UVERBS_METHOD_SRQ_CREATE, + UVERBS_METHOD_SRQ_DESTROY, +}; + enum uverbs_methods_cq { UVERBS_METHOD_CQ_CREATE, UVERBS_METHOD_CQ_DESTROY, -- cgit v1.2.3-59-g8ed1b From ef3bc084a8ed461e3d1f82481f47dacb96596f8f Mon Sep 17 00:00:00 2001 From: Yishai Hadas Date: Tue, 19 May 2020 10:27:10 +0300 Subject: IB/uverbs: Introduce create/destroy WQ commands over ioctl Introduce create/destroy WQ commands over the ioctl interface to let it be extended to get an asynchronous event FD. Link: https://lore.kernel.org/r/20200519072711.257271-7-leon@kernel.org Signed-off-by: Yishai Hadas Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/core/Makefile | 3 +- drivers/infiniband/core/rdma_core.h | 1 + drivers/infiniband/core/uverbs_std_types.c | 23 --- drivers/infiniband/core/uverbs_std_types_wq.c | 194 ++++++++++++++++++++++++++ drivers/infiniband/core/uverbs_uapi.c | 1 + include/uapi/rdma/ib_user_ioctl_cmds.h | 25 ++++ 6 files changed, 223 insertions(+), 24 deletions(-) create mode 100644 drivers/infiniband/core/uverbs_std_types_wq.c (limited to 'include/uapi') diff --git a/drivers/infiniband/core/Makefile b/drivers/infiniband/core/Makefile index d7b46a7c07fd..96c0a4b5af18 100644 --- a/drivers/infiniband/core/Makefile +++ b/drivers/infiniband/core/Makefile @@ -37,6 +37,7 @@ ib_uverbs-y := uverbs_main.o uverbs_cmd.o uverbs_marshall.o \ uverbs_std_types_mr.o uverbs_std_types_counters.o \ uverbs_uapi.o uverbs_std_types_device.o \ uverbs_std_types_async_fd.o \ - uverbs_std_types_srq.o + uverbs_std_types_srq.o \ + uverbs_std_types_wq.o ib_uverbs-$(CONFIG_INFINIBAND_USER_MEM) += umem.o ib_uverbs-$(CONFIG_INFINIBAND_ON_DEMAND_PAGING) += umem_odp.o diff --git a/drivers/infiniband/core/rdma_core.h b/drivers/infiniband/core/rdma_core.h index d623f911b70b..9e9f2fa04fb9 100644 --- a/drivers/infiniband/core/rdma_core.h +++ b/drivers/infiniband/core/rdma_core.h @@ -160,6 +160,7 @@ extern const struct uapi_definition uverbs_def_obj_flow_action[]; extern const struct uapi_definition uverbs_def_obj_intf[]; extern const struct uapi_definition uverbs_def_obj_mr[]; extern const struct uapi_definition uverbs_def_obj_srq[]; +extern const struct uapi_definition uverbs_def_obj_wq[]; extern const struct uapi_definition uverbs_def_write_intf[]; static inline const struct uverbs_api_write_method * diff --git a/drivers/infiniband/core/uverbs_std_types.c b/drivers/infiniband/core/uverbs_std_types.c index d9b6912eafa8..c328d5194076 100644 --- a/drivers/infiniband/core/uverbs_std_types.c +++ b/drivers/infiniband/core/uverbs_std_types.c @@ -125,23 +125,6 @@ static int uverbs_free_rwq_ind_tbl(struct ib_uobject *uobject, return ret; } -static int uverbs_free_wq(struct ib_uobject *uobject, - enum rdma_remove_reason why, - struct uverbs_attr_bundle *attrs) -{ - struct ib_wq *wq = uobject->object; - struct ib_uwq_object *uwq = - container_of(uobject, struct ib_uwq_object, uevent.uobject); - int ret; - - ret = ib_destroy_wq(wq, &attrs->driver_udata); - if (ib_is_destroy_retryable(ret, why, uobject)) - return ret; - - ib_uverbs_release_uevent(&uwq->uevent); - return ret; -} - static int uverbs_free_xrcd(struct ib_uobject *uobject, enum rdma_remove_reason why, struct uverbs_attr_bundle *attrs) @@ -266,10 +249,6 @@ DECLARE_UVERBS_NAMED_OBJECT( uverbs_free_flow), &UVERBS_METHOD(UVERBS_METHOD_FLOW_DESTROY)); -DECLARE_UVERBS_NAMED_OBJECT( - UVERBS_OBJECT_WQ, - UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_uwq_object), uverbs_free_wq)); - DECLARE_UVERBS_NAMED_METHOD_DESTROY( UVERBS_METHOD_RWQ_IND_TBL_DESTROY, UVERBS_ATTR_IDR(UVERBS_ATTR_DESTROY_RWQ_IND_TBL_HANDLE, @@ -318,8 +297,6 @@ const struct uapi_definition uverbs_def_obj_intf[] = { UAPI_DEF_OBJ_NEEDS_FN(dealloc_mw)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_FLOW, UAPI_DEF_OBJ_NEEDS_FN(destroy_flow)), - UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_WQ, - UAPI_DEF_OBJ_NEEDS_FN(destroy_wq)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED( UVERBS_OBJECT_RWQ_IND_TBL, UAPI_DEF_OBJ_NEEDS_FN(destroy_rwq_ind_table)), diff --git a/drivers/infiniband/core/uverbs_std_types_wq.c b/drivers/infiniband/core/uverbs_std_types_wq.c new file mode 100644 index 000000000000..cad842ede077 --- /dev/null +++ b/drivers/infiniband/core/uverbs_std_types_wq.c @@ -0,0 +1,194 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. + */ + +#include +#include "rdma_core.h" +#include "uverbs.h" + +static int uverbs_free_wq(struct ib_uobject *uobject, + enum rdma_remove_reason why, + struct uverbs_attr_bundle *attrs) +{ + struct ib_wq *wq = uobject->object; + struct ib_uwq_object *uwq = + container_of(uobject, struct ib_uwq_object, uevent.uobject); + int ret; + + ret = ib_destroy_wq(wq, &attrs->driver_udata); + if (ib_is_destroy_retryable(ret, why, uobject)) + return ret; + + ib_uverbs_release_uevent(&uwq->uevent); + return ret; +} + +static int UVERBS_HANDLER(UVERBS_METHOD_WQ_CREATE)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uwq_object *obj = container_of( + uverbs_attr_get_uobject(attrs, UVERBS_ATTR_CREATE_WQ_HANDLE), + typeof(*obj), uevent.uobject); + struct ib_pd *pd = + uverbs_attr_get_obj(attrs, UVERBS_ATTR_CREATE_WQ_PD_HANDLE); + struct ib_cq *cq = + uverbs_attr_get_obj(attrs, UVERBS_ATTR_CREATE_WQ_CQ_HANDLE); + struct ib_wq_init_attr wq_init_attr = {}; + struct ib_wq *wq; + u64 user_handle; + int ret; + + ret = uverbs_get_flags32(&wq_init_attr.create_flags, attrs, + UVERBS_ATTR_CREATE_WQ_FLAGS, + IB_UVERBS_WQ_FLAGS_CVLAN_STRIPPING | + IB_UVERBS_WQ_FLAGS_SCATTER_FCS | + IB_UVERBS_WQ_FLAGS_DELAY_DROP | + IB_UVERBS_WQ_FLAGS_PCI_WRITE_END_PADDING); + if (!ret) + ret = uverbs_copy_from(&wq_init_attr.max_sge, attrs, + UVERBS_ATTR_CREATE_WQ_MAX_SGE); + if (!ret) + ret = uverbs_copy_from(&wq_init_attr.max_wr, attrs, + UVERBS_ATTR_CREATE_WQ_MAX_WR); + if (!ret) + ret = uverbs_copy_from(&user_handle, attrs, + UVERBS_ATTR_CREATE_WQ_USER_HANDLE); + if (!ret) + ret = uverbs_get_const(&wq_init_attr.wq_type, attrs, + UVERBS_ATTR_CREATE_WQ_TYPE); + if (ret) + return ret; + + if (wq_init_attr.wq_type != IB_WQT_RQ) + return -EINVAL; + + obj->uevent.event_file = ib_uverbs_get_async_event(attrs, + UVERBS_ATTR_CREATE_WQ_EVENT_FD); + obj->uevent.uobject.user_handle = user_handle; + INIT_LIST_HEAD(&obj->uevent.event_list); + wq_init_attr.event_handler = ib_uverbs_wq_event_handler; + wq_init_attr.wq_context = attrs->ufile; + wq_init_attr.cq = cq; + + wq = pd->device->ops.create_wq(pd, &wq_init_attr, &attrs->driver_udata); + if (IS_ERR(wq)) { + ret = PTR_ERR(wq); + goto err; + } + + obj->uevent.uobject.object = wq; + wq->wq_type = wq_init_attr.wq_type; + wq->cq = cq; + wq->pd = pd; + wq->device = pd->device; + wq->wq_context = wq_init_attr.wq_context; + atomic_set(&wq->usecnt, 0); + atomic_inc(&pd->usecnt); + atomic_inc(&cq->usecnt); + wq->uobject = obj; + uverbs_finalize_uobj_create(attrs, UVERBS_ATTR_CREATE_WQ_HANDLE); + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_WQ_RESP_MAX_WR, + &wq_init_attr.max_wr, + sizeof(wq_init_attr.max_wr)); + if (ret) + return ret; + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_WQ_RESP_MAX_SGE, + &wq_init_attr.max_sge, + sizeof(wq_init_attr.max_sge)); + if (ret) + return ret; + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_WQ_RESP_WQ_NUM, + &wq->wq_num, + sizeof(wq->wq_num)); + return ret; + +err: + if (obj->uevent.event_file) + uverbs_uobject_put(&obj->uevent.event_file->uobj); + return ret; +}; + +DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_WQ_CREATE, + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_WQ_HANDLE, + UVERBS_OBJECT_WQ, + UVERBS_ACCESS_NEW, + UA_MANDATORY), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_WQ_PD_HANDLE, + UVERBS_OBJECT_PD, + UVERBS_ACCESS_READ, + UA_MANDATORY), + UVERBS_ATTR_CONST_IN(UVERBS_ATTR_CREATE_WQ_TYPE, + enum ib_wq_type, + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_WQ_USER_HANDLE, + UVERBS_ATTR_TYPE(u64), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_WQ_MAX_WR, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_WQ_MAX_SGE, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_FLAGS_IN(UVERBS_ATTR_CREATE_WQ_FLAGS, + enum ib_uverbs_wq_flags, + UA_MANDATORY), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_WQ_CQ_HANDLE, + UVERBS_OBJECT_CQ, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_FD(UVERBS_ATTR_CREATE_WQ_EVENT_FD, + UVERBS_OBJECT_ASYNC_EVENT, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_WQ_RESP_MAX_WR, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_WQ_RESP_MAX_SGE, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_WQ_RESP_WQ_NUM, + UVERBS_ATTR_TYPE(u32), + UA_OPTIONAL), + UVERBS_ATTR_UHW()); + +static int UVERBS_HANDLER(UVERBS_METHOD_WQ_DESTROY)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uobject *uobj = + uverbs_attr_get_uobject(attrs, UVERBS_ATTR_DESTROY_WQ_HANDLE); + struct ib_uwq_object *obj = + container_of(uobj, struct ib_uwq_object, uevent.uobject); + + return uverbs_copy_to(attrs, UVERBS_ATTR_DESTROY_WQ_RESP, + &obj->uevent.events_reported, + sizeof(obj->uevent.events_reported)); +} + +DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_WQ_DESTROY, + UVERBS_ATTR_IDR(UVERBS_ATTR_DESTROY_WQ_HANDLE, + UVERBS_OBJECT_WQ, + UVERBS_ACCESS_DESTROY, + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_DESTROY_WQ_RESP, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY)); + + +DECLARE_UVERBS_NAMED_OBJECT( + UVERBS_OBJECT_WQ, + UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_uwq_object), uverbs_free_wq), + &UVERBS_METHOD(UVERBS_METHOD_WQ_CREATE), + &UVERBS_METHOD(UVERBS_METHOD_WQ_DESTROY) +); + +const struct uapi_definition uverbs_def_obj_wq[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_WQ, + UAPI_DEF_OBJ_NEEDS_FN(destroy_wq)), + {} +}; diff --git a/drivers/infiniband/core/uverbs_uapi.c b/drivers/infiniband/core/uverbs_uapi.c index 3f5627954fe7..0ec8cf86ecfa 100644 --- a/drivers/infiniband/core/uverbs_uapi.c +++ b/drivers/infiniband/core/uverbs_uapi.c @@ -635,6 +635,7 @@ static const struct uapi_definition uverbs_core_api[] = { UAPI_DEF_CHAIN(uverbs_def_obj_intf), UAPI_DEF_CHAIN(uverbs_def_obj_mr), UAPI_DEF_CHAIN(uverbs_def_obj_srq), + UAPI_DEF_CHAIN(uverbs_def_obj_wq), UAPI_DEF_CHAIN(uverbs_def_write_intf), {}, }; diff --git a/include/uapi/rdma/ib_user_ioctl_cmds.h b/include/uapi/rdma/ib_user_ioctl_cmds.h index c07af46ff04c..381b17889d20 100644 --- a/include/uapi/rdma/ib_user_ioctl_cmds.h +++ b/include/uapi/rdma/ib_user_ioctl_cmds.h @@ -153,6 +153,31 @@ enum uverbs_methods_cq { UVERBS_METHOD_CQ_DESTROY, }; +enum uverbs_attrs_create_wq_cmd_attr_ids { + UVERBS_ATTR_CREATE_WQ_HANDLE, + UVERBS_ATTR_CREATE_WQ_PD_HANDLE, + UVERBS_ATTR_CREATE_WQ_CQ_HANDLE, + UVERBS_ATTR_CREATE_WQ_USER_HANDLE, + UVERBS_ATTR_CREATE_WQ_TYPE, + UVERBS_ATTR_CREATE_WQ_EVENT_FD, + UVERBS_ATTR_CREATE_WQ_MAX_WR, + UVERBS_ATTR_CREATE_WQ_MAX_SGE, + UVERBS_ATTR_CREATE_WQ_FLAGS, + UVERBS_ATTR_CREATE_WQ_RESP_MAX_WR, + UVERBS_ATTR_CREATE_WQ_RESP_MAX_SGE, + UVERBS_ATTR_CREATE_WQ_RESP_WQ_NUM, +}; + +enum uverbs_attrs_destroy_wq_cmd_attr_ids { + UVERBS_ATTR_DESTROY_WQ_HANDLE, + UVERBS_ATTR_DESTROY_WQ_RESP, +}; + +enum uverbs_methods_wq { + UVERBS_METHOD_WQ_CREATE, + UVERBS_METHOD_WQ_DESTROY, +}; + enum uverbs_methods_actions_flow_action_ops { UVERBS_METHOD_FLOW_ACTION_ESP_CREATE, UVERBS_METHOD_FLOW_ACTION_DESTROY, -- cgit v1.2.3-59-g8ed1b From 6d1e7ba241e990b5c6ba7fdaa03d466f852f3c9e Mon Sep 17 00:00:00 2001 From: Yishai Hadas Date: Tue, 19 May 2020 10:27:11 +0300 Subject: IB/uverbs: Introduce create/destroy QP commands over ioctl Introduce create/destroy QP commands over the ioctl interface to let it be extended to get an asynchronous event FD. Link: https://lore.kernel.org/r/20200519072711.257271-8-leon@kernel.org Signed-off-by: Yishai Hadas Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/core/Makefile | 3 +- drivers/infiniband/core/rdma_core.h | 1 + drivers/infiniband/core/uverbs_std_types.c | 40 --- drivers/infiniband/core/uverbs_std_types_qp.c | 401 ++++++++++++++++++++++++++ drivers/infiniband/core/uverbs_uapi.c | 1 + include/uapi/rdma/ib_user_ioctl_cmds.h | 28 ++ include/uapi/rdma/ib_user_ioctl_verbs.h | 9 + 7 files changed, 442 insertions(+), 41 deletions(-) create mode 100644 drivers/infiniband/core/uverbs_std_types_qp.c (limited to 'include/uapi') diff --git a/drivers/infiniband/core/Makefile b/drivers/infiniband/core/Makefile index 96c0a4b5af18..63c1591223ac 100644 --- a/drivers/infiniband/core/Makefile +++ b/drivers/infiniband/core/Makefile @@ -38,6 +38,7 @@ ib_uverbs-y := uverbs_main.o uverbs_cmd.o uverbs_marshall.o \ uverbs_uapi.o uverbs_std_types_device.o \ uverbs_std_types_async_fd.o \ uverbs_std_types_srq.o \ - uverbs_std_types_wq.o + uverbs_std_types_wq.o \ + uverbs_std_types_qp.o ib_uverbs-$(CONFIG_INFINIBAND_USER_MEM) += umem.o ib_uverbs-$(CONFIG_INFINIBAND_ON_DEMAND_PAGING) += umem_odp.o diff --git a/drivers/infiniband/core/rdma_core.h b/drivers/infiniband/core/rdma_core.h index 9e9f2fa04fb9..33706dad6c0f 100644 --- a/drivers/infiniband/core/rdma_core.h +++ b/drivers/infiniband/core/rdma_core.h @@ -159,6 +159,7 @@ extern const struct uapi_definition uverbs_def_obj_dm[]; extern const struct uapi_definition uverbs_def_obj_flow_action[]; extern const struct uapi_definition uverbs_def_obj_intf[]; extern const struct uapi_definition uverbs_def_obj_mr[]; +extern const struct uapi_definition uverbs_def_obj_qp[]; extern const struct uapi_definition uverbs_def_obj_srq[]; extern const struct uapi_definition uverbs_def_obj_wq[]; extern const struct uapi_definition uverbs_def_write_intf[]; diff --git a/drivers/infiniband/core/uverbs_std_types.c b/drivers/infiniband/core/uverbs_std_types.c index c328d5194076..08c39cfb1bd9 100644 --- a/drivers/infiniband/core/uverbs_std_types.c +++ b/drivers/infiniband/core/uverbs_std_types.c @@ -75,40 +75,6 @@ static int uverbs_free_mw(struct ib_uobject *uobject, return uverbs_dealloc_mw((struct ib_mw *)uobject->object); } -static int uverbs_free_qp(struct ib_uobject *uobject, - enum rdma_remove_reason why, - struct uverbs_attr_bundle *attrs) -{ - struct ib_qp *qp = uobject->object; - struct ib_uqp_object *uqp = - container_of(uobject, struct ib_uqp_object, uevent.uobject); - int ret; - - /* - * If this is a user triggered destroy then do not allow destruction - * until the user cleans up all the mcast bindings. Unlike in other - * places we forcibly clean up the mcast attachments for !DESTROY - * because the mcast attaches are not ubojects and will not be - * destroyed by anything else during cleanup processing. - */ - if (why == RDMA_REMOVE_DESTROY) { - if (!list_empty(&uqp->mcast_list)) - return -EBUSY; - } else if (qp == qp->real_qp) { - ib_uverbs_detach_umcast(qp, uqp); - } - - ret = ib_destroy_qp_user(qp, &attrs->driver_udata); - if (ib_is_destroy_retryable(ret, why, uobject)) - return ret; - - if (uqp->uxrcd) - atomic_dec(&uqp->uxrcd->refcnt); - - ib_uverbs_release_uevent(&uqp->uevent); - return ret; -} - static int uverbs_free_rwq_ind_tbl(struct ib_uobject *uobject, enum rdma_remove_reason why, struct uverbs_attr_bundle *attrs) @@ -210,10 +176,6 @@ DECLARE_UVERBS_NAMED_OBJECT( "[infinibandevent]", O_RDONLY)); -DECLARE_UVERBS_NAMED_OBJECT( - UVERBS_OBJECT_QP, - UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_uqp_object), uverbs_free_qp)); - DECLARE_UVERBS_NAMED_METHOD_DESTROY( UVERBS_METHOD_MW_DESTROY, UVERBS_ATTR_IDR(UVERBS_ATTR_DESTROY_MW_HANDLE, @@ -289,8 +251,6 @@ const struct uapi_definition uverbs_def_obj_intf[] = { UAPI_DEF_OBJ_NEEDS_FN(dealloc_pd)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_COMP_CHANNEL, UAPI_DEF_OBJ_NEEDS_FN(dealloc_pd)), - UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_QP, - UAPI_DEF_OBJ_NEEDS_FN(destroy_qp)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_AH, UAPI_DEF_OBJ_NEEDS_FN(destroy_ah)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_MW, diff --git a/drivers/infiniband/core/uverbs_std_types_qp.c b/drivers/infiniband/core/uverbs_std_types_qp.c new file mode 100644 index 000000000000..3bf8dcdfe7eb --- /dev/null +++ b/drivers/infiniband/core/uverbs_std_types_qp.c @@ -0,0 +1,401 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. + */ + +#include +#include "rdma_core.h" +#include "uverbs.h" +#include "core_priv.h" + +static int uverbs_free_qp(struct ib_uobject *uobject, + enum rdma_remove_reason why, + struct uverbs_attr_bundle *attrs) +{ + struct ib_qp *qp = uobject->object; + struct ib_uqp_object *uqp = + container_of(uobject, struct ib_uqp_object, uevent.uobject); + int ret; + + /* + * If this is a user triggered destroy then do not allow destruction + * until the user cleans up all the mcast bindings. Unlike in other + * places we forcibly clean up the mcast attachments for !DESTROY + * because the mcast attaches are not ubojects and will not be + * destroyed by anything else during cleanup processing. + */ + if (why == RDMA_REMOVE_DESTROY) { + if (!list_empty(&uqp->mcast_list)) + return -EBUSY; + } else if (qp == qp->real_qp) { + ib_uverbs_detach_umcast(qp, uqp); + } + + ret = ib_destroy_qp_user(qp, &attrs->driver_udata); + if (ib_is_destroy_retryable(ret, why, uobject)) + return ret; + + if (uqp->uxrcd) + atomic_dec(&uqp->uxrcd->refcnt); + + ib_uverbs_release_uevent(&uqp->uevent); + return ret; +} + +static int check_creation_flags(enum ib_qp_type qp_type, + u32 create_flags) +{ + create_flags &= ~IB_UVERBS_QP_CREATE_SQ_SIG_ALL; + + if (!create_flags || qp_type == IB_QPT_DRIVER) + return 0; + + if (qp_type != IB_QPT_RAW_PACKET && qp_type != IB_QPT_UD) + return -EINVAL; + + if ((create_flags & IB_UVERBS_QP_CREATE_SCATTER_FCS || + create_flags & IB_UVERBS_QP_CREATE_CVLAN_STRIPPING) && + qp_type != IB_QPT_RAW_PACKET) + return -EINVAL; + + return 0; +} + +static void set_caps(struct ib_qp_init_attr *attr, + struct ib_uverbs_qp_cap *cap, bool req) +{ + if (req) { + attr->cap.max_send_wr = cap->max_send_wr; + attr->cap.max_recv_wr = cap->max_recv_wr; + attr->cap.max_send_sge = cap->max_send_sge; + attr->cap.max_recv_sge = cap->max_recv_sge; + attr->cap.max_inline_data = cap->max_inline_data; + } else { + cap->max_send_wr = attr->cap.max_send_wr; + cap->max_recv_wr = attr->cap.max_recv_wr; + cap->max_send_sge = attr->cap.max_send_sge; + cap->max_recv_sge = attr->cap.max_recv_sge; + cap->max_inline_data = attr->cap.max_inline_data; + } +} + +static int UVERBS_HANDLER(UVERBS_METHOD_QP_CREATE)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uqp_object *obj = container_of( + uverbs_attr_get_uobject(attrs, UVERBS_ATTR_CREATE_QP_HANDLE), + typeof(*obj), uevent.uobject); + struct ib_qp_init_attr attr = {}; + struct ib_uverbs_qp_cap cap = {}; + struct ib_rwq_ind_table *rwq_ind_tbl = NULL; + struct ib_qp *qp; + struct ib_pd *pd = NULL; + struct ib_srq *srq = NULL; + struct ib_cq *recv_cq = NULL; + struct ib_cq *send_cq = NULL; + struct ib_xrcd *xrcd = NULL; + struct ib_uobject *xrcd_uobj = NULL; + struct ib_device *device; + u64 user_handle; + int ret; + + ret = uverbs_copy_from_or_zero(&cap, attrs, + UVERBS_ATTR_CREATE_QP_CAP); + if (!ret) + ret = uverbs_copy_from(&user_handle, attrs, + UVERBS_ATTR_CREATE_QP_USER_HANDLE); + if (!ret) + ret = uverbs_get_const(&attr.qp_type, attrs, + UVERBS_ATTR_CREATE_QP_TYPE); + if (ret) + return ret; + + switch (attr.qp_type) { + case IB_QPT_XRC_TGT: + if (uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_RECV_CQ_HANDLE) || + uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_SEND_CQ_HANDLE) || + uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_PD_HANDLE) || + uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_IND_TABLE_HANDLE)) + return -EINVAL; + + xrcd_uobj = uverbs_attr_get_uobject(attrs, + UVERBS_ATTR_CREATE_QP_XRCD_HANDLE); + if (IS_ERR(xrcd_uobj)) + return PTR_ERR(xrcd_uobj); + + xrcd = (struct ib_xrcd *)xrcd_uobj->object; + if (!xrcd) + return -EINVAL; + device = xrcd->device; + break; + case IB_UVERBS_QPT_RAW_PACKET: + if (!capable(CAP_NET_RAW)) + return -EPERM; + fallthrough; + case IB_UVERBS_QPT_RC: + case IB_UVERBS_QPT_UC: + case IB_UVERBS_QPT_UD: + case IB_UVERBS_QPT_XRC_INI: + case IB_UVERBS_QPT_DRIVER: + if (uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_XRCD_HANDLE) || + (uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_SRQ_HANDLE) && + attr.qp_type == IB_QPT_XRC_INI)) + return -EINVAL; + + pd = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_QP_PD_HANDLE); + if (IS_ERR(pd)) + return PTR_ERR(pd); + + rwq_ind_tbl = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_QP_IND_TABLE_HANDLE); + if (!IS_ERR(rwq_ind_tbl)) { + if (cap.max_recv_wr || cap.max_recv_sge || + uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_RECV_CQ_HANDLE) || + uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_SRQ_HANDLE)) + return -EINVAL; + + /* send_cq is optinal */ + if (cap.max_send_wr) { + send_cq = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_QP_SEND_CQ_HANDLE); + if (IS_ERR(send_cq)) + return PTR_ERR(send_cq); + } + attr.rwq_ind_tbl = rwq_ind_tbl; + } else { + send_cq = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_QP_SEND_CQ_HANDLE); + if (IS_ERR(send_cq)) + return PTR_ERR(send_cq); + + if (attr.qp_type != IB_QPT_XRC_INI) { + recv_cq = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_QP_RECV_CQ_HANDLE); + if (IS_ERR(recv_cq)) + return PTR_ERR(recv_cq); + } + } + + device = pd->device; + break; + default: + return -EINVAL; + } + + ret = uverbs_get_flags32(&attr.create_flags, attrs, + UVERBS_ATTR_CREATE_QP_FLAGS, + IB_UVERBS_QP_CREATE_BLOCK_MULTICAST_LOOPBACK | + IB_UVERBS_QP_CREATE_SCATTER_FCS | + IB_UVERBS_QP_CREATE_CVLAN_STRIPPING | + IB_UVERBS_QP_CREATE_PCI_WRITE_END_PADDING | + IB_UVERBS_QP_CREATE_SQ_SIG_ALL); + if (ret) + return ret; + + ret = check_creation_flags(attr.qp_type, attr.create_flags); + if (ret) + return ret; + + if (uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_SOURCE_QPN)) { + ret = uverbs_copy_from(&attr.source_qpn, attrs, + UVERBS_ATTR_CREATE_QP_SOURCE_QPN); + if (ret) + return ret; + attr.create_flags |= IB_QP_CREATE_SOURCE_QPN; + } + + srq = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_QP_SRQ_HANDLE); + if (!IS_ERR(srq)) { + if ((srq->srq_type == IB_SRQT_XRC && + attr.qp_type != IB_QPT_XRC_TGT) || + (srq->srq_type != IB_SRQT_XRC && + attr.qp_type == IB_QPT_XRC_TGT)) + return -EINVAL; + attr.srq = srq; + } + + obj->uevent.event_file = ib_uverbs_get_async_event(attrs, + UVERBS_ATTR_CREATE_QP_EVENT_FD); + INIT_LIST_HEAD(&obj->uevent.event_list); + INIT_LIST_HEAD(&obj->mcast_list); + obj->uevent.uobject.user_handle = user_handle; + attr.event_handler = ib_uverbs_qp_event_handler; + attr.send_cq = send_cq; + attr.recv_cq = recv_cq; + attr.xrcd = xrcd; + if (attr.create_flags & IB_UVERBS_QP_CREATE_SQ_SIG_ALL) { + /* This creation bit is uverbs one, need to mask before + * calling drivers. It was added to prevent an extra user attr + * only for that when using ioctl. + */ + attr.create_flags &= ~IB_UVERBS_QP_CREATE_SQ_SIG_ALL; + attr.sq_sig_type = IB_SIGNAL_ALL_WR; + } else { + attr.sq_sig_type = IB_SIGNAL_REQ_WR; + } + + set_caps(&attr, &cap, true); + mutex_init(&obj->mcast_lock); + + if (attr.qp_type == IB_QPT_XRC_TGT) + qp = ib_create_qp(pd, &attr); + else + qp = _ib_create_qp(device, pd, &attr, &attrs->driver_udata, + obj); + + if (IS_ERR(qp)) { + ret = PTR_ERR(qp); + goto err_put; + } + + if (attr.qp_type != IB_QPT_XRC_TGT) { + atomic_inc(&pd->usecnt); + if (attr.send_cq) + atomic_inc(&attr.send_cq->usecnt); + if (attr.recv_cq) + atomic_inc(&attr.recv_cq->usecnt); + if (attr.srq) + atomic_inc(&attr.srq->usecnt); + if (attr.rwq_ind_tbl) + atomic_inc(&attr.rwq_ind_tbl->usecnt); + } else { + obj->uxrcd = container_of(xrcd_uobj, struct ib_uxrcd_object, + uobject); + atomic_inc(&obj->uxrcd->refcnt); + /* It is done in _ib_create_qp for other QP types */ + qp->uobject = obj; + } + + obj->uevent.uobject.object = qp; + uverbs_finalize_uobj_create(attrs, UVERBS_ATTR_CREATE_QP_HANDLE); + + if (attr.qp_type != IB_QPT_XRC_TGT) { + ret = ib_create_qp_security(qp, device); + if (ret) + return ret; + } + + set_caps(&attr, &cap, false); + ret = uverbs_copy_to_struct_or_zero(attrs, + UVERBS_ATTR_CREATE_QP_RESP_CAP, &cap, + sizeof(cap)); + if (ret) + return ret; + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_QP_RESP_QP_NUM, + &qp->qp_num, + sizeof(qp->qp_num)); + + return ret; +err_put: + if (obj->uevent.event_file) + uverbs_uobject_put(&obj->uevent.event_file->uobj); + return ret; +}; + +DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_QP_CREATE, + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_HANDLE, + UVERBS_OBJECT_QP, + UVERBS_ACCESS_NEW, + UA_MANDATORY), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_XRCD_HANDLE, + UVERBS_OBJECT_XRCD, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_PD_HANDLE, + UVERBS_OBJECT_PD, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_SRQ_HANDLE, + UVERBS_OBJECT_SRQ, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_SEND_CQ_HANDLE, + UVERBS_OBJECT_CQ, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_RECV_CQ_HANDLE, + UVERBS_OBJECT_CQ, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_IND_TABLE_HANDLE, + UVERBS_OBJECT_RWQ_IND_TBL, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_QP_USER_HANDLE, + UVERBS_ATTR_TYPE(u64), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_QP_CAP, + UVERBS_ATTR_STRUCT(struct ib_uverbs_qp_cap, + max_inline_data), + UA_MANDATORY), + UVERBS_ATTR_CONST_IN(UVERBS_ATTR_CREATE_QP_TYPE, + enum ib_uverbs_qp_type, + UA_MANDATORY), + UVERBS_ATTR_FLAGS_IN(UVERBS_ATTR_CREATE_QP_FLAGS, + enum ib_uverbs_qp_create_flags, + UA_OPTIONAL), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_QP_SOURCE_QPN, + UVERBS_ATTR_TYPE(u32), + UA_OPTIONAL), + UVERBS_ATTR_FD(UVERBS_ATTR_CREATE_QP_EVENT_FD, + UVERBS_OBJECT_ASYNC_EVENT, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_QP_RESP_CAP, + UVERBS_ATTR_STRUCT(struct ib_uverbs_qp_cap, + max_inline_data), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_QP_RESP_QP_NUM, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_UHW()); + +static int UVERBS_HANDLER(UVERBS_METHOD_QP_DESTROY)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uobject *uobj = + uverbs_attr_get_uobject(attrs, UVERBS_ATTR_DESTROY_QP_HANDLE); + struct ib_uqp_object *obj = + container_of(uobj, struct ib_uqp_object, uevent.uobject); + struct ib_uverbs_destroy_qp_resp resp = { + .events_reported = obj->uevent.events_reported + }; + + return uverbs_copy_to(attrs, UVERBS_ATTR_DESTROY_QP_RESP, &resp, + sizeof(resp)); +} + +DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_QP_DESTROY, + UVERBS_ATTR_IDR(UVERBS_ATTR_DESTROY_QP_HANDLE, + UVERBS_OBJECT_QP, + UVERBS_ACCESS_DESTROY, + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_DESTROY_QP_RESP, + UVERBS_ATTR_TYPE(struct ib_uverbs_destroy_qp_resp), + UA_MANDATORY)); + +DECLARE_UVERBS_NAMED_OBJECT( + UVERBS_OBJECT_QP, + UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_uqp_object), uverbs_free_qp), + &UVERBS_METHOD(UVERBS_METHOD_QP_CREATE), + &UVERBS_METHOD(UVERBS_METHOD_QP_DESTROY)); + +const struct uapi_definition uverbs_def_obj_qp[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_QP, + UAPI_DEF_OBJ_NEEDS_FN(destroy_qp)), + {} +}; diff --git a/drivers/infiniband/core/uverbs_uapi.c b/drivers/infiniband/core/uverbs_uapi.c index 0ec8cf86ecfa..5addc8fae3f3 100644 --- a/drivers/infiniband/core/uverbs_uapi.c +++ b/drivers/infiniband/core/uverbs_uapi.c @@ -634,6 +634,7 @@ static const struct uapi_definition uverbs_core_api[] = { UAPI_DEF_CHAIN(uverbs_def_obj_flow_action), UAPI_DEF_CHAIN(uverbs_def_obj_intf), UAPI_DEF_CHAIN(uverbs_def_obj_mr), + UAPI_DEF_CHAIN(uverbs_def_obj_qp), UAPI_DEF_CHAIN(uverbs_def_obj_srq), UAPI_DEF_CHAIN(uverbs_def_obj_wq), UAPI_DEF_CHAIN(uverbs_def_write_intf), diff --git a/include/uapi/rdma/ib_user_ioctl_cmds.h b/include/uapi/rdma/ib_user_ioctl_cmds.h index 381b17889d20..4961d5e858eb 100644 --- a/include/uapi/rdma/ib_user_ioctl_cmds.h +++ b/include/uapi/rdma/ib_user_ioctl_cmds.h @@ -121,6 +121,34 @@ enum uverbs_attrs_destroy_flow_action_esp { UVERBS_ATTR_DESTROY_FLOW_ACTION_HANDLE, }; +enum uverbs_attrs_create_qp_cmd_attr_ids { + UVERBS_ATTR_CREATE_QP_HANDLE, + UVERBS_ATTR_CREATE_QP_XRCD_HANDLE, + UVERBS_ATTR_CREATE_QP_PD_HANDLE, + UVERBS_ATTR_CREATE_QP_SRQ_HANDLE, + UVERBS_ATTR_CREATE_QP_SEND_CQ_HANDLE, + UVERBS_ATTR_CREATE_QP_RECV_CQ_HANDLE, + UVERBS_ATTR_CREATE_QP_IND_TABLE_HANDLE, + UVERBS_ATTR_CREATE_QP_USER_HANDLE, + UVERBS_ATTR_CREATE_QP_CAP, + UVERBS_ATTR_CREATE_QP_TYPE, + UVERBS_ATTR_CREATE_QP_FLAGS, + UVERBS_ATTR_CREATE_QP_SOURCE_QPN, + UVERBS_ATTR_CREATE_QP_EVENT_FD, + UVERBS_ATTR_CREATE_QP_RESP_CAP, + UVERBS_ATTR_CREATE_QP_RESP_QP_NUM, +}; + +enum uverbs_attrs_destroy_qp_cmd_attr_ids { + UVERBS_ATTR_DESTROY_QP_HANDLE, + UVERBS_ATTR_DESTROY_QP_RESP, +}; + +enum uverbs_methods_qp { + UVERBS_METHOD_QP_CREATE, + UVERBS_METHOD_QP_DESTROY, +}; + enum uverbs_attrs_create_srq_cmd_attr_ids { UVERBS_ATTR_CREATE_SRQ_HANDLE, UVERBS_ATTR_CREATE_SRQ_PD_HANDLE, diff --git a/include/uapi/rdma/ib_user_ioctl_verbs.h b/include/uapi/rdma/ib_user_ioctl_verbs.h index b1662dfe86a6..5debab45ebcb 100644 --- a/include/uapi/rdma/ib_user_ioctl_verbs.h +++ b/include/uapi/rdma/ib_user_ioctl_verbs.h @@ -96,6 +96,7 @@ enum ib_uverbs_qp_create_flags { IB_UVERBS_QP_CREATE_SCATTER_FCS = 1 << 8, IB_UVERBS_QP_CREATE_CVLAN_STRIPPING = 1 << 9, IB_UVERBS_QP_CREATE_PCI_WRITE_END_PADDING = 1 << 11, + IB_UVERBS_QP_CREATE_SQ_SIG_ALL = 1 << 12, }; enum ib_uverbs_query_port_cap_flags { @@ -219,6 +220,14 @@ struct ib_uverbs_query_port_resp_ex { __u8 reserved[6]; }; +struct ib_uverbs_qp_cap { + __u32 max_send_wr; + __u32 max_recv_wr; + __u32 max_send_sge; + __u32 max_recv_sge; + __u32 max_inline_data; +}; + enum rdma_driver_id { RDMA_DRIVER_UNKNOWN, RDMA_DRIVER_MLX5, -- cgit v1.2.3-59-g8ed1b From 34e2ab57a911f8b32b22580d11a02f0b79108245 Mon Sep 17 00:00:00 2001 From: Leon Romanovsky Date: Tue, 26 May 2020 13:33:00 +0300 Subject: RDMA/ucma: Extend ucma_connect to receive ECE parameters Active side of CMID initiates connection through librdmacm's rdma_connect() and kernel's ucma_connect(). Extend UCMA interface to handle those new parameters. Link: https://lore.kernel.org/r/20200526103304.196371-3-leon@kernel.org Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/core/cma.c | 21 +++++++++++++++++++++ drivers/infiniband/core/cma_priv.h | 1 + drivers/infiniband/core/ucma.c | 14 +++++++++++--- include/rdma/rdma_cm.h | 3 +++ include/uapi/rdma/rdma_user_cm.h | 6 ++++++ 5 files changed, 42 insertions(+), 3 deletions(-) (limited to 'include/uapi') diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index 432eec472164..e81b8a523a3e 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -4036,6 +4036,27 @@ err: } EXPORT_SYMBOL(rdma_connect); +/** + * rdma_connect_ece - Initiate an active connection request with ECE data. + * @id: Connection identifier to connect. + * @conn_param: Connection information used for connected QPs. + * @ece: ECE parameters + * + * See rdma_connect() explanation. + */ +int rdma_connect_ece(struct rdma_cm_id *id, struct rdma_conn_param *conn_param, + struct rdma_ucm_ece *ece) +{ + struct rdma_id_private *id_priv = + container_of(id, struct rdma_id_private, id); + + id_priv->ece.vendor_id = ece->vendor_id; + id_priv->ece.attr_mod = ece->attr_mod; + + return rdma_connect(id, conn_param); +} +EXPORT_SYMBOL(rdma_connect_ece); + static int cma_accept_ib(struct rdma_id_private *id_priv, struct rdma_conn_param *conn_param) { diff --git a/drivers/infiniband/core/cma_priv.h b/drivers/infiniband/core/cma_priv.h index 5edcf44a9307..caece96ebcf5 100644 --- a/drivers/infiniband/core/cma_priv.h +++ b/drivers/infiniband/core/cma_priv.h @@ -95,6 +95,7 @@ struct rdma_id_private { * Internal to RDMA/core, don't use in the drivers */ struct rdma_restrack_entry res; + struct rdma_ucm_ece ece; }; #if IS_ENABLED(CONFIG_INFINIBAND_ADDR_TRANS_CONFIGFS) diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c index 06127c800a49..7cbb63690241 100644 --- a/drivers/infiniband/core/ucma.c +++ b/drivers/infiniband/core/ucma.c @@ -1072,12 +1072,15 @@ static void ucma_copy_conn_param(struct rdma_cm_id *id, static ssize_t ucma_connect(struct ucma_file *file, const char __user *inbuf, int in_len, int out_len) { - struct rdma_ucm_connect cmd; struct rdma_conn_param conn_param; + struct rdma_ucm_ece ece = {}; + struct rdma_ucm_connect cmd; struct ucma_context *ctx; + size_t in_size; int ret; - if (copy_from_user(&cmd, inbuf, sizeof(cmd))) + in_size = min_t(size_t, in_len, sizeof(cmd)); + if (copy_from_user(&cmd, inbuf, in_size)) return -EFAULT; if (!cmd.conn_param.valid) @@ -1088,8 +1091,13 @@ static ssize_t ucma_connect(struct ucma_file *file, const char __user *inbuf, return PTR_ERR(ctx); ucma_copy_conn_param(ctx->cm_id, &conn_param, &cmd.conn_param); + if (offsetofend(typeof(cmd), ece) <= in_size) { + ece.vendor_id = cmd.ece.vendor_id; + ece.attr_mod = cmd.ece.attr_mod; + } + mutex_lock(&ctx->mutex); - ret = rdma_connect(ctx->cm_id, &conn_param); + ret = rdma_connect_ece(ctx->cm_id, &conn_param, &ece); mutex_unlock(&ctx->mutex); ucma_put_ctx(ctx); return ret; diff --git a/include/rdma/rdma_cm.h b/include/rdma/rdma_cm.h index ea8e794785ed..4e2975eb3643 100644 --- a/include/rdma/rdma_cm.h +++ b/include/rdma/rdma_cm.h @@ -264,6 +264,9 @@ int rdma_init_qp_attr(struct rdma_cm_id *id, struct ib_qp_attr *qp_attr, */ int rdma_connect(struct rdma_cm_id *id, struct rdma_conn_param *conn_param); +int rdma_connect_ece(struct rdma_cm_id *id, struct rdma_conn_param *conn_param, + struct rdma_ucm_ece *ece); + /** * rdma_listen - This function is called by the passive side to * listen for incoming connection requests. diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index 1bb6e75d254b..c1409dd7225f 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -210,10 +210,16 @@ struct rdma_ucm_ud_param { __u8 reserved[7]; }; +struct rdma_ucm_ece { + __u32 vendor_id; + __u32 attr_mod; +}; + struct rdma_ucm_connect { struct rdma_ucm_conn_param conn_param; __u32 id; __u32 reserved; + struct rdma_ucm_ece ece; }; struct rdma_ucm_listen { -- cgit v1.2.3-59-g8ed1b From 93531ee7b9d1313227d2b4f354989895e8d57b72 Mon Sep 17 00:00:00 2001 From: Leon Romanovsky Date: Tue, 26 May 2020 13:33:01 +0300 Subject: RDMA/ucma: Deliver ECE parameters through UCMA events Passive side of CMID connection receives ECE request through REQ message and needs to respond with relevant REP message which will be forwarded to active side. The UCMA events interface is responsible for such communication with the user space (librdmacm). Extend it to provide ECE wire data. Link: https://lore.kernel.org/r/20200526103304.196371-4-leon@kernel.org Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/core/ucma.c | 6 +++++- include/rdma/rdma_cm.h | 1 + include/uapi/rdma/rdma_user_cm.h | 1 + 3 files changed, 7 insertions(+), 1 deletion(-) (limited to 'include/uapi') diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c index 7cbb63690241..3e5268cfa164 100644 --- a/drivers/infiniband/core/ucma.c +++ b/drivers/infiniband/core/ucma.c @@ -360,6 +360,9 @@ static int ucma_event_handler(struct rdma_cm_id *cm_id, ucma_copy_conn_event(&uevent->resp.param.conn, &event->param.conn); + uevent->resp.ece.vendor_id = event->ece.vendor_id; + uevent->resp.ece.attr_mod = event->ece.attr_mod; + if (event->event == RDMA_CM_EVENT_CONNECT_REQUEST) { if (!ctx->backlog) { ret = -ENOMEM; @@ -404,7 +407,8 @@ static ssize_t ucma_get_event(struct ucma_file *file, const char __user *inbuf, * Old 32 bit user space does not send the 4 byte padding in the * reserved field. We don't care, allow it to keep working. */ - if (out_len < sizeof(uevent->resp) - sizeof(uevent->resp.reserved)) + if (out_len < sizeof(uevent->resp) - sizeof(uevent->resp.reserved) - + sizeof(uevent->resp.ece)) return -ENOSPC; if (copy_from_user(&cmd, inbuf, sizeof(cmd))) diff --git a/include/rdma/rdma_cm.h b/include/rdma/rdma_cm.h index 4e2975eb3643..418590c9a9e8 100644 --- a/include/rdma/rdma_cm.h +++ b/include/rdma/rdma_cm.h @@ -111,6 +111,7 @@ struct rdma_cm_event { struct rdma_conn_param conn; struct rdma_ud_param ud; } param; + struct rdma_ucm_ece ece; }; struct rdma_cm_id; diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index c1409dd7225f..19c5c3f74af9 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -297,6 +297,7 @@ struct rdma_ucm_event_resp { struct rdma_ucm_ud_param ud; } param; __u32 reserved; + struct rdma_ucm_ece ece; }; /* Option levels */ -- cgit v1.2.3-59-g8ed1b From 0cb15372a615a9835893f43e86ae45399eb63996 Mon Sep 17 00:00:00 2001 From: Leon Romanovsky Date: Tue, 26 May 2020 13:33:03 +0300 Subject: RDMA/cma: Connect ECE to rdma_accept The rdma_accept() is called by both passive and active sides of CMID connection to mark readiness to start data transfer. For passive side, this is called explicitly, for active side, it is called implicitly while receiving REP message. Provide ECE data to rdma_accept function needed for passive side to send that REP message. Link: https://lore.kernel.org/r/20200526103304.196371-6-leon@kernel.org Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/core/cma.c | 19 +++++++++++++++++++ drivers/infiniband/core/ucma.c | 14 +++++++++++--- include/rdma/rdma_cm.h | 3 +++ include/uapi/rdma/rdma_user_cm.h | 1 + 4 files changed, 34 insertions(+), 3 deletions(-) (limited to 'include/uapi') diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index f554a371f4fa..d449afe5557b 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -4090,6 +4090,8 @@ static int cma_accept_ib(struct rdma_id_private *id_priv, rep.flow_control = conn_param->flow_control; rep.rnr_retry_count = min_t(u8, 7, conn_param->rnr_retry_count); rep.srq = id_priv->srq ? 1 : 0; + rep.ece.vendor_id = id_priv->ece.vendor_id; + rep.ece.attr_mod = id_priv->ece.attr_mod; trace_cm_send_rep(id_priv); ret = ib_send_cm_rep(id_priv->cm_id.ib, &rep); @@ -4137,7 +4139,11 @@ static int cma_send_sidr_rep(struct rdma_id_private *id_priv, return ret; rep.qp_num = id_priv->qp_num; rep.qkey = id_priv->qkey; + + rep.ece.vendor_id = id_priv->ece.vendor_id; + rep.ece.attr_mod = id_priv->ece.attr_mod; } + rep.private_data = private_data; rep.private_data_len = private_data_len; @@ -4195,6 +4201,19 @@ reject: } EXPORT_SYMBOL(__rdma_accept); +int __rdma_accept_ece(struct rdma_cm_id *id, struct rdma_conn_param *conn_param, + const char *caller, struct rdma_ucm_ece *ece) +{ + struct rdma_id_private *id_priv = + container_of(id, struct rdma_id_private, id); + + id_priv->ece.vendor_id = ece->vendor_id; + id_priv->ece.attr_mod = ece->attr_mod; + + return __rdma_accept(id, conn_param, caller); +} +EXPORT_SYMBOL(__rdma_accept_ece); + int rdma_notify(struct rdma_cm_id *id, enum ib_event_type event) { struct rdma_id_private *id_priv; diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c index 3e5268cfa164..6b27b210b890 100644 --- a/drivers/infiniband/core/ucma.c +++ b/drivers/infiniband/core/ucma.c @@ -1135,28 +1135,36 @@ static ssize_t ucma_accept(struct ucma_file *file, const char __user *inbuf, { struct rdma_ucm_accept cmd; struct rdma_conn_param conn_param; + struct rdma_ucm_ece ece = {}; struct ucma_context *ctx; + size_t in_size; int ret; - if (copy_from_user(&cmd, inbuf, sizeof(cmd))) + in_size = min_t(size_t, in_len, sizeof(cmd)); + if (copy_from_user(&cmd, inbuf, in_size)) return -EFAULT; ctx = ucma_get_ctx_dev(file, cmd.id); if (IS_ERR(ctx)) return PTR_ERR(ctx); + if (offsetofend(typeof(cmd), ece) <= in_size) { + ece.vendor_id = cmd.ece.vendor_id; + ece.attr_mod = cmd.ece.attr_mod; + } + if (cmd.conn_param.valid) { ucma_copy_conn_param(ctx->cm_id, &conn_param, &cmd.conn_param); mutex_lock(&file->mut); mutex_lock(&ctx->mutex); - ret = __rdma_accept(ctx->cm_id, &conn_param, NULL); + ret = __rdma_accept_ece(ctx->cm_id, &conn_param, NULL, &ece); mutex_unlock(&ctx->mutex); if (!ret) ctx->uid = cmd.uid; mutex_unlock(&file->mut); } else { mutex_lock(&ctx->mutex); - ret = __rdma_accept(ctx->cm_id, NULL, NULL); + ret = __rdma_accept_ece(ctx->cm_id, NULL, NULL, &ece); mutex_unlock(&ctx->mutex); } ucma_put_ctx(ctx); diff --git a/include/rdma/rdma_cm.h b/include/rdma/rdma_cm.h index 418590c9a9e8..7ac91677660f 100644 --- a/include/rdma/rdma_cm.h +++ b/include/rdma/rdma_cm.h @@ -280,6 +280,9 @@ int rdma_listen(struct rdma_cm_id *id, int backlog); int __rdma_accept(struct rdma_cm_id *id, struct rdma_conn_param *conn_param, const char *caller); +int __rdma_accept_ece(struct rdma_cm_id *id, struct rdma_conn_param *conn_param, + const char *caller, struct rdma_ucm_ece *ece); + /** * rdma_accept - Called to accept a connection request or response. * @id: Connection identifier associated with the request. diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index 19c5c3f74af9..6b883dde7064 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -232,6 +232,7 @@ struct rdma_ucm_accept { struct rdma_ucm_conn_param conn_param; __u32 id; __u32 reserved; + struct rdma_ucm_ece ece; }; struct rdma_ucm_reject { -- cgit v1.2.3-59-g8ed1b From 8094ba0ace7f6cd1e31ea8b151fba3594cadfa9a Mon Sep 17 00:00:00 2001 From: Leon Romanovsky Date: Tue, 26 May 2020 13:33:04 +0300 Subject: RDMA/cma: Provide ECE reject reason IBTA declares "vendor option not supported" reject reason in REJ messages if passive side doesn't want to accept proposed ECE options. Due to the fact that ECE is managed by userspace, there is a need to let users to provide such rejected reason. Link: https://lore.kernel.org/r/20200526103304.196371-7-leon@kernel.org Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/core/cma.c | 9 ++++----- drivers/infiniband/core/ucma.c | 15 ++++++++++++++- drivers/infiniband/ulp/isert/ib_isert.c | 5 +++-- drivers/infiniband/ulp/rtrs/rtrs-srv.c | 3 ++- drivers/infiniband/ulp/srpt/ib_srpt.c | 3 ++- drivers/nvme/target/rdma.c | 4 +++- include/rdma/rdma_cm.h | 2 +- include/uapi/rdma/rdma_user_cm.h | 3 ++- net/rds/ib_cm.c | 4 +++- 9 files changed, 34 insertions(+), 14 deletions(-) (limited to 'include/uapi') diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index d449afe5557b..8026ee56546a 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -4196,7 +4196,7 @@ int __rdma_accept(struct rdma_cm_id *id, struct rdma_conn_param *conn_param, return 0; reject: cma_modify_qp_err(id_priv); - rdma_reject(id, NULL, 0); + rdma_reject(id, NULL, 0, IB_CM_REJ_CONSUMER_DEFINED); return ret; } EXPORT_SYMBOL(__rdma_accept); @@ -4236,7 +4236,7 @@ int rdma_notify(struct rdma_cm_id *id, enum ib_event_type event) EXPORT_SYMBOL(rdma_notify); int rdma_reject(struct rdma_cm_id *id, const void *private_data, - u8 private_data_len) + u8 private_data_len, u8 reason) { struct rdma_id_private *id_priv; int ret; @@ -4251,9 +4251,8 @@ int rdma_reject(struct rdma_cm_id *id, const void *private_data, private_data, private_data_len); } else { trace_cm_send_rej(id_priv); - ret = ib_send_cm_rej(id_priv->cm_id.ib, - IB_CM_REJ_CONSUMER_DEFINED, NULL, - 0, private_data, private_data_len); + ret = ib_send_cm_rej(id_priv->cm_id.ib, reason, NULL, 0, + private_data, private_data_len); } } else if (rdma_cap_iw_cm(id->device, id->port_num)) { ret = iw_cm_reject(id_priv->cm_id.iw, diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c index 6b27b210b890..5b87eee8ccc8 100644 --- a/drivers/infiniband/core/ucma.c +++ b/drivers/infiniband/core/ucma.c @@ -52,6 +52,7 @@ #include #include #include +#include #include #include "core_priv.h" @@ -1181,12 +1182,24 @@ static ssize_t ucma_reject(struct ucma_file *file, const char __user *inbuf, if (copy_from_user(&cmd, inbuf, sizeof(cmd))) return -EFAULT; + if (!cmd.reason) + cmd.reason = IB_CM_REJ_CONSUMER_DEFINED; + + switch (cmd.reason) { + case IB_CM_REJ_CONSUMER_DEFINED: + case IB_CM_REJ_VENDOR_OPTION_NOT_SUPPORTED: + break; + default: + return -EINVAL; + } + ctx = ucma_get_ctx_dev(file, cmd.id); if (IS_ERR(ctx)) return PTR_ERR(ctx); mutex_lock(&ctx->mutex); - ret = rdma_reject(ctx->cm_id, cmd.private_data, cmd.private_data_len); + ret = rdma_reject(ctx->cm_id, cmd.private_data, cmd.private_data_len, + cmd.reason); mutex_unlock(&ctx->mutex); ucma_put_ctx(ctx); return ret; diff --git a/drivers/infiniband/ulp/isert/ib_isert.c b/drivers/infiniband/ulp/isert/ib_isert.c index a1a035270cab..b7df38ee8ae0 100644 --- a/drivers/infiniband/ulp/isert/ib_isert.c +++ b/drivers/infiniband/ulp/isert/ib_isert.c @@ -15,6 +15,7 @@ #include #include #include +#include #include #include #include @@ -502,7 +503,7 @@ isert_connect_request(struct rdma_cm_id *cma_id, struct rdma_cm_event *event) if (!np->enabled) { spin_unlock_bh(&np->np_thread_lock); isert_dbg("iscsi_np is not enabled, reject connect request\n"); - return rdma_reject(cma_id, NULL, 0); + return rdma_reject(cma_id, NULL, 0, IB_CM_REJ_CONSUMER_DEFINED); } spin_unlock_bh(&np->np_thread_lock); @@ -553,7 +554,7 @@ out_rsp_dma_map: isert_free_login_buf(isert_conn); out: kfree(isert_conn); - rdma_reject(cma_id, NULL, 0); + rdma_reject(cma_id, NULL, 0, IB_CM_REJ_CONSUMER_DEFINED); return ret; } diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv.c b/drivers/infiniband/ulp/rtrs/rtrs-srv.c index 5ef8988ee75b..0d9241f5d9e6 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-srv.c +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv.c @@ -15,6 +15,7 @@ #include "rtrs-srv.h" #include "rtrs-log.h" +#include MODULE_DESCRIPTION("RDMA Transport Server"); MODULE_LICENSE("GPL"); @@ -1576,7 +1577,7 @@ static int rtrs_rdma_do_reject(struct rdma_cm_id *cm_id, int errno) .errno = cpu_to_le16(errno), }; - err = rdma_reject(cm_id, &msg, sizeof(msg)); + err = rdma_reject(cm_id, &msg, sizeof(msg), IB_CM_REJ_CONSUMER_DEFINED); if (err) pr_err("rdma_reject(), err: %d\n", err); diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c index a294630f2100..cdc8c239d6c0 100644 --- a/drivers/infiniband/ulp/srpt/ib_srpt.c +++ b/drivers/infiniband/ulp/srpt/ib_srpt.c @@ -2497,7 +2497,8 @@ reject: SRP_BUF_FORMAT_INDIRECT); if (rdma_cm_id) - rdma_reject(rdma_cm_id, rej, sizeof(*rej)); + rdma_reject(rdma_cm_id, rej, sizeof(*rej), + IB_CM_REJ_CONSUMER_DEFINED); else ib_send_cm_rej(ib_cm_id, IB_CM_REJ_CONSUMER_DEFINED, NULL, 0, rej, sizeof(*rej)); diff --git a/drivers/nvme/target/rdma.c b/drivers/nvme/target/rdma.c index fd47de0e4e4e..55aaf03a9580 100644 --- a/drivers/nvme/target/rdma.c +++ b/drivers/nvme/target/rdma.c @@ -20,6 +20,7 @@ #include #include #include +#include #include #include "nvmet.h" @@ -1138,7 +1139,8 @@ static int nvmet_rdma_cm_reject(struct rdma_cm_id *cm_id, rej.recfmt = cpu_to_le16(NVME_RDMA_CM_FMT_1_0); rej.sts = cpu_to_le16(status); - return rdma_reject(cm_id, (void *)&rej, sizeof(rej)); + return rdma_reject(cm_id, (void *)&rej, sizeof(rej), + IB_CM_REJ_CONSUMER_DEFINED); } static struct nvmet_rdma_queue * diff --git a/include/rdma/rdma_cm.h b/include/rdma/rdma_cm.h index 7ac91677660f..939d7abe026f 100644 --- a/include/rdma/rdma_cm.h +++ b/include/rdma/rdma_cm.h @@ -320,7 +320,7 @@ int rdma_notify(struct rdma_cm_id *id, enum ib_event_type event); * rdma_reject - Called to reject a connection request or response. */ int rdma_reject(struct rdma_cm_id *id, const void *private_data, - u8 private_data_len); + u8 private_data_len, u8 reason); /** * rdma_disconnect - This function disconnects the associated QP and diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index 6b883dde7064..ed5a514305c1 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -238,7 +238,8 @@ struct rdma_ucm_accept { struct rdma_ucm_reject { __u32 id; __u8 private_data_len; - __u8 reserved[3]; + __u8 reason; + __u8 reserved[2]; __u8 private_data[RDMA_MAX_PRIVATE_DATA]; }; diff --git a/net/rds/ib_cm.c b/net/rds/ib_cm.c index c71f4328d138..0fec4171564e 100644 --- a/net/rds/ib_cm.c +++ b/net/rds/ib_cm.c @@ -37,6 +37,7 @@ #include #include #include +#include #include "rds_single_path.h" #include "rds.h" @@ -927,7 +928,8 @@ out: if (conn) mutex_unlock(&conn->c_cm_lock); if (err) - rdma_reject(cm_id, &err, sizeof(int)); + rdma_reject(cm_id, &err, sizeof(int), + IB_CM_REJ_CONSUMER_DEFINED); return destroy; } -- cgit v1.2.3-59-g8ed1b From 3e09a427ae7ac347e08dca5ffac64c902860d675 Mon Sep 17 00:00:00 2001 From: Leon Romanovsky Date: Tue, 26 May 2020 14:54:34 +0300 Subject: RDMA/mlx5: Get ECE options from FW during create QP Supported ECE options are returned from FW in the create_qp phase and zero means that field is not valid. Such default value allows us to reuse reserved field without worries about comp_mask. Update create QP API to return ECE options. Link: https://lore.kernel.org/r/20200526115440.205922-3-leon@kernel.org Reviewed-by: Mark Zhang Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/hw/mlx5/qp.c | 16 +++++++++++----- drivers/infiniband/hw/mlx5/qp.h | 4 ++-- drivers/infiniband/hw/mlx5/qpc.c | 8 ++++---- include/uapi/rdma/mlx5-abi.h | 2 +- 4 files changed, 18 insertions(+), 12 deletions(-) (limited to 'include/uapi') diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index 2e28752e8cd2..be7289c480f7 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -1842,6 +1842,7 @@ static int create_xrc_tgt_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, struct ib_qp_init_attr *attr = params->attr; u32 uidx = params->uidx; struct mlx5_ib_resources *devr = &dev->devr; + u32 out[MLX5_ST_SZ_DW(create_qp_out)] = {}; int inlen = MLX5_ST_SZ_BYTES(create_qp_in); struct mlx5_core_dev *mdev = dev->mdev; struct mlx5_ib_qp_base *base; @@ -1894,13 +1895,14 @@ static int create_xrc_tgt_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, } base = &qp->trans_qp.base; - err = mlx5_core_create_qp(dev, &base->mqp, in, inlen); + err = mlx5_qpc_create_qp(dev, &base->mqp, in, inlen, out); kvfree(in); if (err) return err; base->container_mibqp = qp; base->mqp.event = mlx5_ib_qp_event; + params->resp.ece_options = MLX5_GET(create_qp_out, out, ece); spin_lock_irqsave(&dev->reset_flow_resource_lock, flags); list_add_tail(&qp->qps_list, &dev->qp_list); @@ -1916,6 +1918,7 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, { struct ib_qp_init_attr *init_attr = params->attr; struct mlx5_ib_create_qp *ucmd = params->ucmd; + u32 out[MLX5_ST_SZ_DW(create_qp_out)] = {}; struct ib_udata *udata = params->udata; u32 uidx = params->uidx; struct mlx5_ib_resources *devr = &dev->devr; @@ -2065,7 +2068,7 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, err = create_raw_packet_qp(dev, qp, in, inlen, pd, udata, ¶ms->resp); } else - err = mlx5_core_create_qp(dev, &base->mqp, in, inlen); + err = mlx5_qpc_create_qp(dev, &base->mqp, in, inlen, out); kvfree(in); if (err) @@ -2073,6 +2076,7 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, base->container_mibqp = qp; base->mqp.event = mlx5_ib_qp_event; + params->resp.ece_options = MLX5_GET(create_qp_out, out, ece); get_cqs(qp->type, init_attr->send_cq, init_attr->recv_cq, &send_cq, &recv_cq); @@ -2105,6 +2109,7 @@ static int create_kernel_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, struct ib_qp_init_attr *attr = params->attr; u32 uidx = params->uidx; struct mlx5_ib_resources *devr = &dev->devr; + u32 out[MLX5_ST_SZ_DW(create_qp_out)] = {}; int inlen = MLX5_ST_SZ_BYTES(create_qp_in); struct mlx5_core_dev *mdev = dev->mdev; struct mlx5_ib_cq *send_cq; @@ -2195,7 +2200,7 @@ static int create_kernel_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, if (qp->flags & IB_QP_CREATE_IPOIB_UD_LSO) MLX5_SET(qpc, qpc, ulp_stateless_offload_mode, 1); - err = mlx5_core_create_qp(dev, &base->mqp, in, inlen); + err = mlx5_qpc_create_qp(dev, &base->mqp, in, inlen, out); kvfree(in); if (err) goto err_create; @@ -2779,12 +2784,13 @@ out: qp->ibqp.qp_num = qp->trans_qp.base.mqp.qpn; mlx5_ib_dbg(dev, - "QP type %d, ib qpn 0x%X, mlx qpn 0x%x, rcqn 0x%x, scqn 0x%x\n", + "QP type %d, ib qpn 0x%X, mlx qpn 0x%x, rcqn 0x%x, scqn 0x%x, ece 0x%x\n", qp->type, qp->ibqp.qp_num, qp->trans_qp.base.mqp.qpn, params->attr->recv_cq ? to_mcq(params->attr->recv_cq)->mcq.cqn : -1, params->attr->send_cq ? to_mcq(params->attr->send_cq)->mcq.cqn : - -1); + -1, + params->resp.ece_options); return 0; } diff --git a/drivers/infiniband/hw/mlx5/qp.h b/drivers/infiniband/hw/mlx5/qp.h index ad9d76e3e18a..795c21f88962 100644 --- a/drivers/infiniband/hw/mlx5/qp.h +++ b/drivers/infiniband/hw/mlx5/qp.h @@ -13,8 +13,8 @@ void mlx5_cleanup_qp_table(struct mlx5_ib_dev *dev); int mlx5_core_create_dct(struct mlx5_ib_dev *dev, struct mlx5_core_dct *qp, u32 *in, int inlen, u32 *out, int outlen); -int mlx5_core_create_qp(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp, - u32 *in, int inlen); +int mlx5_qpc_create_qp(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp, + u32 *in, int inlen, u32 *out); int mlx5_core_qp_modify(struct mlx5_ib_dev *dev, u16 opcode, u32 opt_param_mask, void *qpc, struct mlx5_core_qp *qp); int mlx5_core_destroy_qp(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp); diff --git a/drivers/infiniband/hw/mlx5/qpc.c b/drivers/infiniband/hw/mlx5/qpc.c index ea62735042f0..69c80859a6ee 100644 --- a/drivers/infiniband/hw/mlx5/qpc.c +++ b/drivers/infiniband/hw/mlx5/qpc.c @@ -236,16 +236,16 @@ err_cmd: return err; } -int mlx5_core_create_qp(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp, - u32 *in, int inlen) +int mlx5_qpc_create_qp(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp, + u32 *in, int inlen, u32 *out) { - u32 out[MLX5_ST_SZ_DW(create_qp_out)] = {}; u32 din[MLX5_ST_SZ_DW(destroy_qp_in)] = {}; int err; MLX5_SET(create_qp_in, in, opcode, MLX5_CMD_OP_CREATE_QP); - err = mlx5_cmd_exec(dev->mdev, in, inlen, out, sizeof(out)); + err = mlx5_cmd_exec(dev->mdev, in, inlen, out, + MLX5_ST_SZ_BYTES(create_qp_out)); if (err) return err; diff --git a/include/uapi/rdma/mlx5-abi.h b/include/uapi/rdma/mlx5-abi.h index df1cc3641bda..106fbb3bec6a 100644 --- a/include/uapi/rdma/mlx5-abi.h +++ b/include/uapi/rdma/mlx5-abi.h @@ -371,7 +371,7 @@ enum mlx5_ib_create_qp_resp_mask { struct mlx5_ib_create_qp_resp { __u32 bfreg_index; - __u32 reserved; + __u32 ece_options; __u32 comp_mask; __u32 tirn; __u32 tisn; -- cgit v1.2.3-59-g8ed1b From e383085c24255821e79d3c2aa6302d804b6a1c48 Mon Sep 17 00:00:00 2001 From: Leon Romanovsky Date: Tue, 26 May 2020 14:54:35 +0300 Subject: RDMA/mlx5: Set ECE options during QP create Allow users to ask creation of QPs with specific ECE options. Such early set even before RDMA-CM connection is established is useful if user knows exactly which option he needs. Link: https://lore.kernel.org/r/20200526115440.205922-4-leon@kernel.org Reviewed-by: Mark Zhang Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/hw/mlx5/qp.c | 75 ++++++++++++++++++++++++++++++++++++----- include/uapi/rdma/mlx5-abi.h | 2 ++ 2 files changed, 68 insertions(+), 9 deletions(-) (limited to 'include/uapi') diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index be7289c480f7..eb70eb371b4b 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -1552,6 +1552,7 @@ struct mlx5_create_qp_params { struct ib_udata *udata; size_t inlen; size_t outlen; + size_t ucmd_size; void *ucmd; u8 is_rss_raw : 1; struct ib_qp_init_attr *attr; @@ -1839,6 +1840,7 @@ static int get_atomic_mode(struct mlx5_ib_dev *dev, static int create_xrc_tgt_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, struct mlx5_create_qp_params *params) { + struct mlx5_ib_create_qp *ucmd = params->ucmd; struct ib_qp_init_attr *attr = params->attr; u32 uidx = params->uidx; struct mlx5_ib_resources *devr = &dev->devr; @@ -1860,6 +1862,8 @@ static int create_xrc_tgt_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, if (!in) return -ENOMEM; + if (MLX5_CAP_GEN(mdev, ece_support)) + MLX5_SET(create_qp_in, in, ece, ucmd->ece_options); qpc = MLX5_ADDR_OF(create_qp_in, in, qpc); MLX5_SET(qpc, qpc, st, MLX5_QP_ST_XRC); @@ -1974,6 +1978,8 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, if (is_sqp(init_attr->qp_type)) qp->port = init_attr->port_num; + if (MLX5_CAP_GEN(mdev, ece_support)) + MLX5_SET(create_qp_in, in, ece, ucmd->ece_options); qpc = MLX5_ADDR_OF(create_qp_in, in, qpc); MLX5_SET(qpc, qpc, st, mlx5_st); @@ -2709,19 +2715,22 @@ static int process_udata_size(struct mlx5_ib_dev *dev, struct mlx5_create_qp_params *params) { size_t ucmd = sizeof(struct mlx5_ib_create_qp); - struct ib_qp_init_attr *attr = params->attr; struct ib_udata *udata = params->udata; size_t outlen = udata->outlen; size_t inlen = udata->inlen; params->outlen = min(outlen, sizeof(struct mlx5_ib_create_qp_resp)); - if (attr->qp_type == IB_QPT_DRIVER) { - params->inlen = (inlen < ucmd) ? 0 : ucmd; - goto out; - } - + params->ucmd_size = ucmd; if (!params->is_rss_raw) { - params->inlen = ucmd; + /* User has old rdma-core, which doesn't support ECE */ + size_t min_inlen = + offsetof(struct mlx5_ib_create_qp, ece_options); + + /* + * We will check in check_ucmd_data() that user + * cleared everything after inlen. + */ + params->inlen = (inlen < min_inlen) ? 0 : min(inlen, ucmd); goto out; } @@ -2733,13 +2742,14 @@ static int process_udata_size(struct mlx5_ib_dev *dev, return -EINVAL; ucmd = sizeof(struct mlx5_ib_create_qp_rss); + params->ucmd_size = ucmd; if (inlen > ucmd && !ib_is_udata_cleared(udata, ucmd, inlen - ucmd)) return -EINVAL; params->inlen = min(ucmd, inlen); out: if (!params->inlen) - mlx5_ib_dbg(dev, "udata is too small or not cleared\n"); + mlx5_ib_dbg(dev, "udata is too small\n"); return (params->inlen) ? 0 : -EINVAL; } @@ -2855,6 +2865,49 @@ static int mlx5_ib_destroy_dct(struct mlx5_ib_qp *mqp) return 0; } +static int check_ucmd_data(struct mlx5_ib_dev *dev, + struct mlx5_create_qp_params *params) +{ + struct ib_qp_init_attr *attr = params->attr; + struct ib_udata *udata = params->udata; + size_t size, last; + int ret; + + if (params->is_rss_raw) + /* + * These QPs don't have "reserved" field in their + * create_qp input struct, so their data is always valid. + */ + last = sizeof(struct mlx5_ib_create_qp_rss); + else + /* IB_QPT_RAW_PACKET and IB_QPT_DRIVER don't have ECE data */ + switch (attr->qp_type) { + case IB_QPT_DRIVER: + case IB_QPT_RAW_PACKET: + last = offsetof(struct mlx5_ib_create_qp, ece_options); + break; + default: + last = offsetof(struct mlx5_ib_create_qp, reserved); + } + + if (udata->inlen <= last) + return 0; + + /* + * User provides different create_qp structures based on the + * flow and we need to know if he cleared memory after our + * struct create_qp ends. + */ + size = udata->inlen - last; + ret = ib_is_udata_cleared(params->udata, last, size); + if (!ret) + mlx5_ib_dbg( + dev, + "udata is not cleared, inlen = %lu, ucmd = %lu, last = %lu, size = %lu\n", + udata->inlen, params->ucmd_size, last, size); + return ret ? 0 : -EINVAL; +} + struct ib_qp *mlx5_ib_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attr, struct ib_udata *udata) { @@ -2888,7 +2941,11 @@ struct ib_qp *mlx5_ib_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attr, if (err) return ERR_PTR(err); - params.ucmd = kzalloc(params.inlen, GFP_KERNEL); + err = check_ucmd_data(dev, ¶ms); + if (err) + return ERR_PTR(err); + + params.ucmd = kzalloc(params.ucmd_size, GFP_KERNEL); if (!params.ucmd) return ERR_PTR(-ENOMEM); diff --git a/include/uapi/rdma/mlx5-abi.h b/include/uapi/rdma/mlx5-abi.h index 106fbb3bec6a..bc9d9e3cb369 100644 --- a/include/uapi/rdma/mlx5-abi.h +++ b/include/uapi/rdma/mlx5-abi.h @@ -322,6 +322,8 @@ struct mlx5_ib_create_qp { __aligned_u64 sq_buf_addr; __aligned_u64 access_key; }; + __u32 ece_options; + __u32 reserved; }; /* RX Hash function flags */ -- cgit v1.2.3-59-g8ed1b From 5f62a521ff20e0b47a8d33421334bd245d6714ff Mon Sep 17 00:00:00 2001 From: Leon Romanovsky Date: Tue, 26 May 2020 14:54:39 +0300 Subject: RDMA/mlx5: Set ECE options during modify QP The most common way to set ECE option will be during modify QP command in INIT2RTR, RTR2RTS and RTS2RTS stages, so update mlx5 to support it. The new bit in the comp_mask is needed to mark that kernel supports ECE and can receive data instead of "reserved" field in the struct mlx5_ib_modify_qp. Link: https://lore.kernel.org/r/20200526115440.205922-8-leon@kernel.org Reviewed-by: Mark Zhang Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/hw/mlx5/main.c | 3 +++ drivers/infiniband/hw/mlx5/qp.c | 21 ++++++++++++--------- drivers/infiniband/hw/mlx5/qp.h | 2 +- drivers/infiniband/hw/mlx5/qpc.c | 11 +++++++---- include/uapi/rdma/mlx5-abi.h | 3 ++- 5 files changed, 25 insertions(+), 15 deletions(-) (limited to 'include/uapi') diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index 623d7898ae6d..6557c8339161 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -1959,6 +1959,9 @@ uar_done: resp.response_length += sizeof(resp.dump_fill_mkey); } + if (MLX5_CAP_GEN(dev->mdev, ece_support)) + resp.comp_mask |= MLX5_IB_ALLOC_UCONTEXT_RESP_MASK_ECE; + err = ib_copy_to_udata(udata, &resp, resp.response_length); if (err) goto out_mdev; diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index a24176a8ec83..bfa0f7e43e3b 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -2353,7 +2353,7 @@ static void destroy_qp_common(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, if (qp->ibqp.qp_type != IB_QPT_RAW_PACKET && !(qp->flags & IB_QP_CREATE_SOURCE_QPN)) { err = mlx5_core_qp_modify(dev, MLX5_CMD_OP_2RST_QP, 0, - NULL, &base->mqp); + NULL, &base->mqp, NULL); } else { struct mlx5_modify_raw_qp_param raw_qp_param = { .operation = MLX5_CMD_OP_2RST_QP @@ -3978,7 +3978,10 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, err = modify_raw_packet_qp(dev, qp, &raw_qp_param, tx_affinity); } else { - err = mlx5_core_qp_modify(dev, op, optpar, qpc, &base->mqp); + u32 ece = MLX5_CAP_GEN(dev->mdev, ece_support) ? + ucmd->ece_options : 0; + err = mlx5_core_qp_modify(dev, op, optpar, qpc, &base->mqp, + &ece); } if (err) @@ -4131,7 +4134,6 @@ static int mlx5_ib_modify_dct(struct ib_qp *ibqp, struct ib_qp_attr *attr, set_id = mlx5_ib_get_counters_id(dev, attr->port_num - 1); MLX5_SET(dctc, dctc, counter_set_id, set_id); - } else if (cur_state == IB_QPS_INIT && new_state == IB_QPS_RTR) { struct mlx5_ib_modify_qp_resp resp = {}; u32 out[MLX5_ST_SZ_DW(create_dct_out)] = {0}; @@ -4182,7 +4184,6 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, struct mlx5_ib_modify_qp ucmd = {}; enum ib_qp_type qp_type; enum ib_qp_state cur_state, new_state; - size_t required_cmd_sz; int err = -EINVAL; int port; @@ -4190,9 +4191,7 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, return -ENOSYS; if (udata && udata->inlen) { - required_cmd_sz = offsetof(typeof(ucmd), reserved) + - sizeof(ucmd.reserved); - if (udata->inlen < required_cmd_sz) + if (udata->inlen < offsetofend(typeof(ucmd), ece_options)) return -EINVAL; if (udata->inlen > sizeof(ucmd) && @@ -4205,10 +4204,10 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, return -EFAULT; if (ucmd.comp_mask || - memchr_inv(&ucmd.reserved, 0, sizeof(ucmd.reserved)) || memchr_inv(&ucmd.burst_info.reserved, 0, sizeof(ucmd.burst_info.reserved))) return -EOPNOTSUPP; + } if (unlikely(ibqp->qp_type == IB_QPT_GSI)) @@ -4217,8 +4216,12 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, qp_type = (unlikely(ibqp->qp_type == MLX5_IB_QPT_HW_GSI)) ? IB_QPT_GSI : qp->type; - if (qp_type == MLX5_IB_QPT_DCT) + if (qp_type == MLX5_IB_QPT_DCT) { + if (memchr_inv(&ucmd.ece_options, 0, sizeof(ucmd.ece_options))) + return -EOPNOTSUPP; + return mlx5_ib_modify_dct(ibqp, attr, attr_mask, udata); + } mutex_lock(&qp->mutex); diff --git a/drivers/infiniband/hw/mlx5/qp.h b/drivers/infiniband/hw/mlx5/qp.h index 795c21f88962..82ea2b94dfa6 100644 --- a/drivers/infiniband/hw/mlx5/qp.h +++ b/drivers/infiniband/hw/mlx5/qp.h @@ -16,7 +16,7 @@ int mlx5_core_create_dct(struct mlx5_ib_dev *dev, struct mlx5_core_dct *qp, int mlx5_qpc_create_qp(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp, u32 *in, int inlen, u32 *out); int mlx5_core_qp_modify(struct mlx5_ib_dev *dev, u16 opcode, u32 opt_param_mask, - void *qpc, struct mlx5_core_qp *qp); + void *qpc, struct mlx5_core_qp *qp, u32 *ece); int mlx5_core_destroy_qp(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp); int mlx5_core_destroy_dct(struct mlx5_ib_dev *dev, struct mlx5_core_dct *dct); int mlx5_core_qp_query(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp, diff --git a/drivers/infiniband/hw/mlx5/qpc.c b/drivers/infiniband/hw/mlx5/qpc.c index 69c80859a6ee..d61bc1a88925 100644 --- a/drivers/infiniband/hw/mlx5/qpc.c +++ b/drivers/infiniband/hw/mlx5/qpc.c @@ -343,7 +343,7 @@ static void mbox_free(struct mbox_info *mbox) static int modify_qp_mbox_alloc(struct mlx5_core_dev *dev, u16 opcode, int qpn, u32 opt_param_mask, void *qpc, - struct mbox_info *mbox, u16 uid) + struct mbox_info *mbox, u16 uid, u32 ece) { mbox->out = NULL; mbox->in = NULL; @@ -391,18 +391,21 @@ static int modify_qp_mbox_alloc(struct mlx5_core_dev *dev, u16 opcode, int qpn, return -ENOMEM; MOD_QP_IN_SET_QPC(init2rtr_qp, mbox->in, opcode, qpn, opt_param_mask, qpc, uid); + MLX5_SET(init2rtr_qp_in, mbox->in, ece, ece); break; case MLX5_CMD_OP_RTR2RTS_QP: if (MBOX_ALLOC(mbox, rtr2rts_qp)) return -ENOMEM; MOD_QP_IN_SET_QPC(rtr2rts_qp, mbox->in, opcode, qpn, opt_param_mask, qpc, uid); + MLX5_SET(rtr2rts_qp_in, mbox->in, ece, ece); break; case MLX5_CMD_OP_RTS2RTS_QP: if (MBOX_ALLOC(mbox, rts2rts_qp)) return -ENOMEM; MOD_QP_IN_SET_QPC(rts2rts_qp, mbox->in, opcode, qpn, opt_param_mask, qpc, uid); + MLX5_SET(rts2rts_qp_in, mbox->in, ece, ece); break; case MLX5_CMD_OP_SQERR2RTS_QP: if (MBOX_ALLOC(mbox, sqerr2rts_qp)) @@ -423,13 +426,13 @@ static int modify_qp_mbox_alloc(struct mlx5_core_dev *dev, u16 opcode, int qpn, } int mlx5_core_qp_modify(struct mlx5_ib_dev *dev, u16 opcode, u32 opt_param_mask, - void *qpc, struct mlx5_core_qp *qp) + void *qpc, struct mlx5_core_qp *qp, u32 *ece) { struct mbox_info mbox; int err; - err = modify_qp_mbox_alloc(dev->mdev, opcode, qp->qpn, - opt_param_mask, qpc, &mbox, qp->uid); + err = modify_qp_mbox_alloc(dev->mdev, opcode, qp->qpn, opt_param_mask, + qpc, &mbox, qp->uid, (ece) ? *ece : 0); if (err) return err; diff --git a/include/uapi/rdma/mlx5-abi.h b/include/uapi/rdma/mlx5-abi.h index bc9d9e3cb369..24e29a678177 100644 --- a/include/uapi/rdma/mlx5-abi.h +++ b/include/uapi/rdma/mlx5-abi.h @@ -100,6 +100,7 @@ struct mlx5_ib_alloc_ucontext_req_v2 { enum mlx5_ib_alloc_ucontext_resp_mask { MLX5_IB_ALLOC_UCONTEXT_RESP_MASK_CORE_CLOCK_OFFSET = 1UL << 0, MLX5_IB_ALLOC_UCONTEXT_RESP_MASK_DUMP_FILL_MKEY = 1UL << 1, + MLX5_IB_ALLOC_UCONTEXT_RESP_MASK_ECE = 1UL << 2, }; enum mlx5_user_cmds_supp_uhw { @@ -422,7 +423,7 @@ struct mlx5_ib_burst_info { struct mlx5_ib_modify_qp { __u32 comp_mask; struct mlx5_ib_burst_info burst_info; - __u32 reserved; + __u32 ece_options; }; struct mlx5_ib_modify_qp_resp { -- cgit v1.2.3-59-g8ed1b From 50aec2c3135efd985291adc2e4d1278d52b03de9 Mon Sep 17 00:00:00 2001 From: Leon Romanovsky Date: Tue, 26 May 2020 14:54:40 +0300 Subject: RDMA/mlx5: Return ECE data after modify QP After users sets the ECE option, FW will return the agreed/supported bits through an output structures of modify QP stages for regular QPs or through create QP for the DCT. Link: https://lore.kernel.org/r/20200526115440.205922-9-leon@kernel.org Reviewed-by: Mark Zhang Signed-off-by: Leon Romanovsky Signed-off-by: Jason Gunthorpe --- drivers/infiniband/hw/mlx5/qp.c | 25 +++++++++++++++++++++---- drivers/infiniband/hw/mlx5/qpc.c | 25 +++++++++++++++++++++++++ include/uapi/rdma/mlx5-abi.h | 2 ++ 3 files changed, 48 insertions(+), 4 deletions(-) (limited to 'include/uapi') diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index bfa0f7e43e3b..1988a0375696 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -3708,6 +3708,7 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, enum ib_qp_state cur_state, enum ib_qp_state new_state, const struct mlx5_ib_modify_qp *ucmd, + struct mlx5_ib_modify_qp_resp *resp, struct ib_udata *udata) { static const u16 optab[MLX5_QP_NUM_STATE][MLX5_QP_NUM_STATE] = { @@ -3978,10 +3979,15 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, err = modify_raw_packet_qp(dev, qp, &raw_qp_param, tx_affinity); } else { - u32 ece = MLX5_CAP_GEN(dev->mdev, ece_support) ? - ucmd->ece_options : 0; + if (udata) { + /* For the kernel flows, the resp will stay zero */ + resp->ece_options = + MLX5_CAP_GEN(dev->mdev, ece_support) ? + ucmd->ece_options : 0; + resp->response_length = sizeof(*resp); + } err = mlx5_core_qp_modify(dev, op, optpar, qpc, &base->mqp, - &ece); + &resp->ece_options); } if (err) @@ -4180,6 +4186,7 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, int attr_mask, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(ibqp->device); + struct mlx5_ib_modify_qp_resp resp = {}; struct mlx5_ib_qp *qp = to_mqp(ibqp); struct mlx5_ib_modify_qp ucmd = {}; enum ib_qp_type qp_type; @@ -4292,7 +4299,17 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, } err = __mlx5_ib_modify_qp(ibqp, attr, attr_mask, cur_state, - new_state, &ucmd, udata); + new_state, &ucmd, &resp, udata); + + /* resp.response_length is set in ECE supported flows only */ + if (!err && resp.response_length && + udata->outlen >= resp.response_length) + /* + * We don't check return value of the function below + * on purpose, because it is unclear how to unwind the + * error flow after QP was modified to the new state. + */ + ib_copy_to_udata(udata, &resp, resp.response_length); out: mutex_unlock(&qp->mutex); diff --git a/drivers/infiniband/hw/mlx5/qpc.c b/drivers/infiniband/hw/mlx5/qpc.c index d61bc1a88925..c19d91d6dce8 100644 --- a/drivers/infiniband/hw/mlx5/qpc.c +++ b/drivers/infiniband/hw/mlx5/qpc.c @@ -341,6 +341,27 @@ static void mbox_free(struct mbox_info *mbox) kfree(mbox->out); } +static int get_ece_from_mbox(void *out, u16 opcode) +{ + int ece = 0; + + switch (opcode) { + case MLX5_CMD_OP_INIT2RTR_QP: + ece = MLX5_GET(init2rtr_qp_out, out, ece); + break; + case MLX5_CMD_OP_RTR2RTS_QP: + ece = MLX5_GET(rtr2rts_qp_out, out, ece); + break; + case MLX5_CMD_OP_RTS2RTS_QP: + ece = MLX5_GET(rts2rts_qp_out, out, ece); + break; + default: + break; + } + + return ece; +} + static int modify_qp_mbox_alloc(struct mlx5_core_dev *dev, u16 opcode, int qpn, u32 opt_param_mask, void *qpc, struct mbox_info *mbox, u16 uid, u32 ece) @@ -438,6 +459,10 @@ int mlx5_core_qp_modify(struct mlx5_ib_dev *dev, u16 opcode, u32 opt_param_mask, err = mlx5_cmd_exec(dev->mdev, mbox.in, mbox.inlen, mbox.out, mbox.outlen); + + if (ece) + *ece = get_ece_from_mbox(mbox.out, opcode); + mbox_free(&mbox); return err; } diff --git a/include/uapi/rdma/mlx5-abi.h b/include/uapi/rdma/mlx5-abi.h index 24e29a678177..27905a0268c9 100644 --- a/include/uapi/rdma/mlx5-abi.h +++ b/include/uapi/rdma/mlx5-abi.h @@ -429,6 +429,8 @@ struct mlx5_ib_modify_qp { struct mlx5_ib_modify_qp_resp { __u32 response_length; __u32 dctn; + __u32 ece_options; + __u32 reserved; }; struct mlx5_ib_create_wq_resp { -- cgit v1.2.3-59-g8ed1b